*
Report to the United States Senate
Nuclear Accident
and Recovery at
Three Mile Island
A Special Investigation
Subcommittee on Nuclear Regulation
for the
Senate Committee on Environment & Public Works
From the collection of the
o PreTinger
ibrary
San Francisco, California
2007
o ? on f ress 1 COMMITTEE PRINT
2d Session J
NUCLEAR ACCIDENT AND RECOVERY AT
THREE MILE ISLAND
A REPORT
PREPARED BY THE
SUBCOMMITTEE ON NUCLEAR REGULATION
FOR THE
COMMITTEE ON ENVIRONMENT AND
PUBLIC WORKS
U.S. SENATE
JUNE 1980
SERIAL NO. 96-14
Printed for the use of the Committee on Environment and Public Works
U.S. GOVERNMENT PRINTING OFFICE
M-058 O WASHINGTON : 1980
COMMITTEE ON ENVIRONMENT AND PUBLIC WORKS
JENNINGS RANDOLPH, West Virginia, Chairman
MIKE GRAVEL, Alaska ROBERT T. STAFFORD, Vermont
LLOYD M. BENTSEN, Texas HOWARD H. BAKER, JR., Tennessee
QUENTIN N. BURDICK, North Dakota PETE V. DOMENICI, New Mexico
JOHN C. CULVER, Iowa JOHN H. CHAFEE, Rhode Island
GARY HART, Colorado ALAN K. SIMPSON, Wyoming
DANIEL PATRICK MOYNIHAN, New York LARRY PRESSLER, South Dakota
GEORGE J. MITCHELL, Maine
JOHN W. YAGO, Jr., Staff Director
BAILEY GUARD, Minority Staff Director
SUBCOMMITTEE ON NUCLEAR REGULATION
GARY HART, Colorado, Chairman
JENNINGS RANDOLPH, West Virginia ALAN K. SIMPSON, Wyoming
JOHN C. CULVER, Iowa HOWARD H. BAKER, JR., Tennessee
DANIEL PATRICK MOYNIHAN, New York PETE V. DOMENICI, New Mexico
(H)
LETTER OF TRANSMITTAL
UXITED STATES SEXATE,
COMMITTEE OX ExVIROXMEXT AXD PfBLIC WORKS,
Washington, D.C., June 23, 1980.
Honorable WALTER F. MOXDALE,
Pi' *',<! nt of the Senate.
Washington. D.C.
DEAR MR. PRESIDENT : AVe transmit herewith the report "Nuclear Accident and Recovery at Three
Mile Island." the product of a special investigation carried out by the Subcommittee on Nuclear Regula-
tion for the Committee on Environment and Public Works.
This report was developed by the Subcommittee under the Committee's standing jurisdiction over
non-military environmental control and regulation of nuclear energy. We believe it will make a contribu-
tion to congressional and public understanding of the Three Mile Island accident.
Truly.
JEXXIXGS RAXDOLPH.
Chairman,
ROBERT T. STAFFORD.
Ranking Minority Member.
LETTER OF SUBMITTAL
UNITED STATES SENATE,
COMMITTEE ox ENVIRONMENT AND PUBLIC WORKS.
Washington. D.C... June <7, 1980.
THE HOXORABLE JENNINGS RANDOLPH,
Chairman.
THE HONORABLE ROBERT T. STAFFORD,
Ranking Minority Member,
X finite Committee on Environment and Public Works,
Washington, D.C.
DEAR JENNINGS AND BOB: We are pleased to transmit to you the report, "Nuclear Accident and
Recovery at Three Mile Island," which is the result of the Special Investigation conducted for the Com-
mittee on Environment and Public Works by the Subcommittee on Nuclear Regulation.
On June 21, 1979, the Senate provided the Committee additional funds for the investigation and
for a series of related policy studies on major issues arising out of the accident.
The one-year task of the Special Investigation will be completed by the end of June when the policy
studies are transmitted to the Committee.
We believe the report makes a valuable contribution to Congressional and public understanding of
the Three Mile Island accident. The investigation was conducted with a temporary staff that operated
on a non-partisan, unified basis. All findings and conclusions are keyed to supporting facts in the text of
the report, and these facts, in turn, are fully referenced to supporting documents.
The Special Investigation reviewed the work of other principal investigations of the accident, in-
cluding those of the Presidential Commission on the Three Mile Island Accident and the Special Inquiry
of the Nuclear Regulatory Commission. Our report also covers some areas not emphasized in the other
investigations. We provide a detailed accounting of the first day of the accident, of certain pre-accident
conditions that related directly to the accident, and of the cleanup and recovery operations at Three
Mile Island.
By concentrating in greater depth on these areas, we believe that this report relates more directly to
the oversight interests of the Committee and to possible legislative action to be taken by the Committee.
We extend our appreciation for your remarkable support and assistance throughout the term of
the project. Your wise counsel and advice have been most important to the Committee Members and staff.
We believe that the report is consistent with the high standards of the Environment and Public Works
Committee of the U.S. Senate.
We also extend our appreciation to our superb staff which has worked unstintingly and profes-
sionally out of the limelight to produce a report that we all hope will contribute to serious and
thoughtful public discussion of the future direction of nuclear power in our society.
Sincerely,
GABY HART,
Chairman,
Subcommittee on Nuclear Regulation.
ALAN K. SIMPSON,
Ranking Minority Member,
Subcommittee on Nuclear Regulation.
(V)
TABLE OF CONTENTS
Page
Chapter 1: Introduction 1
Staff of the Three Mile Island Special Investigation 5
Chapter 2: Findings And Conclusions 7
I. The Accident
II. Recover}' 19
Chapter 3: How The Plant Works 23
Nuclear vs. Non-Nuclear Plants 2.5
Three Mile Island, Unit 2 25
Chapter 4: I low The Accident Happened: A Mechanical Summary 33
The First Seconds " 35
Steam in the System 36
Core Uncovering 37
A Site, Then General Emergency 37
Strategies to Reach Stability 37
Chapter 5 : Radiation Effects And Monitoring 41
Measuring Radiation 43
Radiation Monitoring at TMI 44
Chapter 6: Prior To The Accident 47
Introduction 49
The Evolution of Unit 2 50
Related Accidents at Other Plants 76
Emergency Response Planning 79
Chapter 7: Accident At Three Mile Island: The First Day 87
Introduction 93
4:00:36 The Beginning 93
A Site Emergency Is Declared 110
A General Emergency Is Declared 112
Stable Conditions Achieved 151
Addenda to Chapter 7 153
Chapter 8: Recovery At Three Mile Island 161
Introduction 163
Technical Aspects of Recovery 164
Financial Aspects of Recovery 190
Social Issues in Recovery 196
Legal and Regulatory Aspects of Recovery 201
Appendix A : Three Mile Island In Perspective : Other Nuclear Accidents 219
Appendix B: Nuclear Regulatory Commission Organization 227
Appendix C: Nuclear Regulatory Commission Reactor Licensing Process 233
Appendix D: Chronology of First-Day Responses 239
Appendix E: Technical Glossary 365
Appendix F: Glossarj* of Organizations 377
References _. 383
PHOTO CREDITS
Allied Fix Service. Inc., pp. 60. 105, 174; Babcock & Wilcox. p. 75 : Metropolitan Edison Co.. pp. 42,
4s. !C). 177. 180. 181 ; Nuclear Regulatory Commission, pp. 2. 8, 55, 88. 92, 133 ; Wide World Photos, p. 162.
(VH)
Chapter 1
Introduction
- .
The Three Mile Island nuclear power plant
Chapter 1
Introduction
The Committee on Environment and Public Works has jurisdiction over all matters relating to
environmental regulation and control of civilian nuclear energy, including the activities of the Nuclear
Regulatory Commission (XRC). This responsibility is carried out through the Subcommittee on Nuclear
Regulation. The Committee was assigned this regulatory jurisdiction after the Joint Committee on
Atomic Energy was abolished in 1977.
On June -21. 1979. the Senate approved S. Res. 171. providing funds for the Committee for a Special
Investigation of the Three Mile Island nuclear accident and for a series of related policy studies on
Federal regulation and control of the nuclear power program. The report accompanying the resolution
>tated that both the investigation and the policy studies were to be completed within one year for the
Committee on Environment and Public Works by the Subcommittee on Nuclear Regulation.
This report on the investigation of the accident at Three Mile Island fulfills part of that assignment.
The policy studies conducted by the Special Investigation staff will be completed by the end of the one
year project. Together, the investigation report and the studies should assist the Congress in exercising
'its responsibility with respect to the Three Mile Island accident and in considering the need for legisla-
tive and administrative changes for more effective regulation and control of commercial nuclear power.
This independent Congressional investigation is consistent with the unique role of the Congress in
the Nation's atomic energy program, a role that dates back to the establishment of the Atomic Energy
Commission in 1946. Congress established the Nuclear Regulatory Commission in 1974, as it had the
earlier Atomic Energy Commission, to be independent of the President and of the Executive Branch
and to keep Congress "fully and currently informed" of its activities. This Congressional involvement
carries with it a responsibility to develop independent findings and to come to independent conclusions
about the facts and implications of Three Mile Island.
All findings and conclusions in the report are keyed to supporting facts in the text. These facts, in
turn, are extensively referenced to supporting documentation. Reference numbers appear in parentheses
in the text : the references themselves are the last section of the report.
This Senate Special Investigation is one of several inquiries into the accident. The inquiries of The
President's Commission on the Accident at Three Mile Island and of the Nuclear Regulatory Commis-
sion Special Inquiry Group have been completed. The Senate investigation of the accident differed in
several respects from those of the Presidential Commission and the NRC. This investigation was kept
small in size and selective in scope, to avoid duplication of the comprehensive approach to the accident
taken by the other inquiries.
The Special Investigation, consistent with an objective stated in the report accompanying the
Senate resolution, examined the work of these other inquiries as part of its independent assessment of
the accident. For example, the Special Investigation staff, in addition to examining original plant
records, reviewed the files of other investigations to explore design and mechanical aspects of the
accident. This permitted the Subcommittee to conduct the investigation and studies on a limited budget
and with a small investigative staff. The General Accounting Office and the Congressional Research
Service also provided assistance.
The Subcommittee on Nuclear Regulation received testimony on the accident from 61 witnesses at
eight hearings. The Special Investigation staff conducted 97 transcribed interviews and conducted
numerous others.
The Subcommittee and the investigation staff heard from a broad spectrum of those involved in
and concerned with the accident, including members and staff of the Nuclear Regulatory Commission,
both at NRC headquarters in the Washington, D.C. area and at the Commission's Pennsylvania regional
office; executives of the parent and operating utilities, of the Three Mile Island plant designer and of
the reactor- vendor ; the Governor and Lieutenant Governor of Pennsylvania and senior State officials ;
local elected officials and interested citizens from communities near the plant; company and government
workers involved in building and licensing the plant and in responding to the accident; members or
staff of two State public utility commissions; members of The President's Commission and investigators
from that inquiry and from the Special Inquiry Group of the NRC: and nuclear advocates, nuclear
critics and concerned citizens.
The Special Investigation also examined thousands of pages of transcripts, depositions and other
documents from the files of the NRC and of the involved companies.
Given the volume of this material, it was necessary and useful to be selective about what was investi-
gated. The Subcommittee concentrated on specific areas that relate to the oversight interests of the
Committee and to possible legislative action by the Congress.
The Subcommittee, for example, chose to focus on the first 24 hours of the accident. In doing so, the
Special Investigation staff reviewed all of the, transcripts of the telephone conversations between the
TMI site and NRC headquarters during the first day the period that is now known to have involved
the greatest instability and uncertainty at the plant. This section of the report presents a detailed narra-
tive of what was known and not known to plant operators and managers, and to NRC and State officials,
and endeavors to account for why actions were and were not taken.
Beyond the uncertain hours of the first day of the accident, the Subcommittee focused on the cleanup
operation at the site of the crippled reactor. More than a year after the accident, only a small portion of
the cleanup work had been accomplished. Because cleanup and recovery at Three Mile Island are an
extension of the accident itself, the Subcommittee explored this timely problem in all of its ramifica-
tions technical, financial, social, legal and regulatory.
In addition to the events of the first day of and the first year after the accident, this report traces
the evolution of the TMI Unit-2 plant from its originally proposed site on the Atlantic coast at Oyster
Creek, New Jersey, to the island in the middle of the Susquehanna River where it was eventually built.
The move to the new site affected subsequent decisions about the design of the plant's control room. Some
elements of the control room design contributed to difficulties encountered by plant operators during the
accident. The report also describes problems encountered during early testing and operation of the plant
that were directly related to the accident.
The findings and conclusions of the investigation appear in Chapter 2 of this report. They are fol-
lowed by three brief introductory chapters, intended for the general reader, describing in non-technical
terms "How the Plant Works," "How the Accident Happened," and "Radiation Effects and Monitoring."
The text of the investigation report itself follows these sections.
Glossaries of technical terms and of governmental and private organizations involved in the accident
and recovery appear at the end of the report as appendices. Other appendices provide a chronology of
the first-day responses to the accident ; review previous nuclear accidents ; and describe the organization
of the NRC and the NRC licensing process.
The policy studies will focus on the adequacy of certain programs begun by the Nuclear Regulatory
Commission and the nuclear industry as a result of lessons learned from Three Mile Island. In particular,
the studies will examine programs to improve
consideration of human factors in nuclear plant design and operation,
the evaluation and dissemination of information on mechanical and operating problems experi-
enced at nuclear power plants,
emergency response to nuclear accidents, and
the industry's new insurance program to cover replacement power costs following an accident.
Finally, a word about the non-partisan nature of this investigation. In keeping with a tradition
of the Committee of a close working relationship between the majority and minority members, a tempo-
rary Special Investigation staff was set up on a unified, non-partisan basis. There were no separate ma-
jority and minority investigation staffs; there are no separate majority and minority views in this report.
This was possible because of close cooperation between the Chairman and the Ranking Minority Mem-
ber, and among all members, of the Nuclear Regulation Subcommittee.
The Subcommittee acknowledges the outstanding work of the Special Investigation staff .
STAFF OF THE THREE MILE ISLAND SPECIAL INVESTIGATION
CODIRECTORS
Paul L. Leventhal and James K. Asselstine
COUNSEL
William G. Ballaine
Steven M. Blush
Jay E. Boudreau
TASK GROUP LEADERS
David D. Carlson
Joan M. Giannelli
INVESTIGATIVE STAFF
David E. Bucher
Carla D'Arista
Katherine W. Kimball
Vivien F. Lee
Mark E. Recktenwald
Monte Simpson
Eoy Squares
OFFICE MANAGER
Joan K. Ramsay
SUPPORT STAFF
Cheryl G. Brown
Irene S. Sarate
Mary Helen Sullivan
EDITOR
Whitney Watriss
PRODUCTION COORDINATOR
Roy Squares
CONSULTANT
Bruce Mann
The Subcommittee acknowledges the contributions made by Jonathan C. Cottin and Drew C. Arena
as members of the Special Investigation staff through December 1979. They did not participate in the
preparation of the final report.
Chapter 2
Findings and Conclusions
Victor Stello of the NRC testifies before the Senate Subcommittee on Nuclear Regulation on the accident
at Three Mile Island
Chapter 2
Findings and Conclusions 1
I. THE ACCIDENT
A. CAUSES OF THE ACCIDENT
1. Malfunctions in plant equipment - initiated
the accident at Three Mile Island, but they alone
did not cause the uncovering of the core or the
severity and duration of the accident. Feedwater
transients such as the one that initiated the March
28 accident occur routinely at nuclear power
plants. They result from a variety of minor equip-
ment malfunctions or from human error such as
experienced at TMI. 3
Routine transient.-; can evolve into serious acci-
dents if complicated bv human factor deficiencies
and other deficiencies in training, in control room
design, in instrumentation and equipment, in
emergency procedures and in plant design. The
psychological stress experienced by plant person-
nel during a crisis is a further complicating factor.
All of these factors can serve to confuse plant
personnel and to render them unable to respond
to a minor accident effectively. At TMI, these fac-
tors caused a minor event to evolve into a serious
accident.
2. Plant operators and managers inappropri-
ately overrode the automatic safety equipment
actions that were the immediate cause of the un-
covering of. and severe damage to, the reactor
core. 4 However, it is inappropriate and unfair sim-
ply to blame these personnel for the Three Mile
Island accident. It should be emphasized that the
utility, the reactor-vendor, the architect-engineer
and the XRC were responsible for deficiencies in
training. 5 in control room design. 6 in instrumen-
tation and equipment, 7 in plant design, 8 and in
emergency procedures. 9 These deficiencies were the
underlying cause of the accident.
Many of these deficiencies resulted from insuffi-
cient attention by the utility, the reactor-vendor,
the architect-engineer and the XRC to human fac-
tors in nuclear plant design and operation. 10 These
human factor problems were l^eyond the control
of the operators on duty during the accident and
were so serious that they had consequences equiv-
alent to those that could be caused solely by major
mechanical failures and design defects.
3. Several major weaknesses in the design of
TMI-2 contributed to the difficulties faced by
plant operators and managers in understanding
plant behavior, in stabilizing the plant, and par-
ticularly hi preventing radiological releases to the
environment. 11 In some cases they involved equip-
ment designed for use in an accident that failed
to fulfill its intended purpose on March 28. 12 In
other cases, design had focused on normal operat-
ing conditions; instrumentation and equipment
needed or useful under the emergency conditions
at TMI had not been provided or were inadequate
to the task. 13 These design weaknesses are of con-
cern because of their possible generic safety
implications.
Design weaknesses in the emergency-related
equipment included :
A system of some 1.200 alarms, of which
several hundred went off in the first minutes.
Operators said they had concluded prior to the
accident that the alarms would provide little, if
any. immediate assistance in diagnosing a major
transient or in assigning priorities to accident con-
ditions. 14 After the accident, operators said the
alarms were "not very helpful" 15 and "got in the
wav." 16
1 The reader may find it useful to read first the introductory chapters. "How the Plant Works," "How the Accident
Happened" and "Radiation Effects and Monitoring" for descriptions of plant systems and explanations of technical
terms.
! Most particularly problems with the condensate polishing system and the failure of the pilot-operated relief ralve
(PORV I . See p. 94 of the text. All pase references in this chapter are to the text.
3 "Staff Report on the Generic Assessment of Feedwater Transients in Pressurized Water Reactors Designed by the
Babcock & Wilcox Company." XRC, NTREG-0560, May 1979.
4 Pp. 93-110. * Pp. 73-76. ' Pp. 56-64. ' Pp. 65-6. 69-71, 94, 96. 99-101. 103. 104. 155. ' Pp. 96. 99-100.
' P. 58. Pp. 56, 60-63. " Pp. 94-96. 99-101. 103-104, 155. H Pp. 96, 99, 103-104. " Pp. 94, 96. 100-101, 106.
113-114116-117. M Pp. 68-70, 99. "P. 99. "P. 99.
51-058 - 80 - 2
A computer printer that was, as anticipated
by the operators, of little help because it failed to
keep pace with the sequence of alarms 17 and be-
came severely backlogged. 18
A radiation monitor that was intended to be
a key indicator of a loss-of-coolant accident
(LOCA) but apparently did not sound on March
28. Prior to the accident it may have been mis-
calibrated, and on the first day it may have become
disabled by the steam and water resulting from
the LOCA."
The failure of the containment building to
seal automatically on initiation of high pressure
injection, resulting in the automatic pumping of
radioactive water from the containment into the
unsealed auxiliary building. 20
Design weaknesses related to equipment that was
needed in the emergency, but was unavailable or
inadequate to the task, included :
The lack of a direct indicator to show whether
the pilot-operated relief valve (PORV) was open
or closed. 21
Indicators of conditions in the reactor coolant
drain tank (pointing to a LOCA) that were not
directly visible to plant operators from the main
console in the control room. 22
The lack of strip chart recorders for reactor
coolant drain tank conditions, without which it
was difficult for operators to reconstruct trends in
the tank's temperature, pressure and water level. 23
The lack of instrumentation to measure water
level in the reactor vessel directly. Instead, opera-
tors had to rely on water level in the pressurizer
as an indirect indicator that proved unreliable
during the accident. 24
The inability to maintain isolation of the con-
tainment building when use of the let-down system
was required to cope with the accident. 28
The inability to seal off the pathways between
the auxiliary building and the environment to pre-
vent releases of radioactivity to the environment
after operators overrode containment isolation in
order to use the let-down system.
Instrumention that was designed only for nor-
mal operating conditions and could not provide
readings for the extreme conditions produced by
the accident. 28 Thus control room personnel could
not monitor those extreme conditions directly. 27
Since these misleading readings influenced actions
taken to control the accident, the limited range of
the instruments was a particularly significant
weakness in plant design.
In addition, as had happened before during
early testing of the plant, the "candy-cane" curve
in the hotlegs trapped steam formed from boiling
of the coolant. This blockage inhibited natural
circulation and contributed to difficulties in under-
standing plant behavior and in stabilizing the
plant.
Had these weaknesses not been present in the
design of the plant, the operators and managers
would have been in a better position to understand
and to respond to the accident.
4. The emergency procedures for Unit 2 were
vague, confusing, incomplete and not fully under-
stood by plant personnel. 28 They did not provide
useful guidance to operators and managers in
identifying and responding to the critical elements
of the accident in the early hours. 29
Better emergency procedures and better under-
standing of them by plant operators and managers
would have facilitated diagnosis and understand-
ing of the plant's behavior. It should be noted,
however, that it is impossible to write emergency
procedures to fit every possible accident sequence.
5. There were several weaknesses in the TMI
operator training program that contributed to the
difficulty control room personnel had in under-
standing and responding to the sequence of events
of the March 28 accident. 30
These weaknesses included :
Limited training in multiple-failure acci-
dents, particularly such prolonged ones as experi-
enced on March 28 at TMI ; 31
Limited training in the basics of nuclear
power plant physics and behavior; 32
Failure to instruct operators on conditions in
which water level in the pressurizer would not be
a reliable indicator of water level in the reactor
vessel. Operators had been directed never to let the
pressurizer fill completely ("go solid") with water
during plant operation. 33 This direction had been
based on the concern that a pressurizer "solid"
with water could limit their ability to control
pressure in the primary system and could result in
damage to the plant. 34
Operators and managers would have been better
prepared to respond to the accident if their train-
injr had been more extensive in these areas.
1 Pp. 65-66, 94-95, 155. " Pp. 100-101.
" Pp. 69-70, 99. 1 Pp. 69-71, 99. " Pp. 103-104. M Pp. 99-101.
23 Pp. 100-101. " P. 96. M P. 96.
26 Examples were the computer and control panel instrumentation used to monitor critical plant parameters, includ-
ing temperatures in the hotlegs of the primary coolant system and temperatures inside the core. The scale for hotleg
temperatures on the control panel went only to 620 F (they reached an estimated 720-820 F during the accident) (see
pp. 106, 113, 132, 142) ; the computer could print out incore temperatures only as high as 700 F (they reached an esti-
mated 4,500 F), see p. 113.
27 Pp. 113, 114. M Pp. 102-104, 154-156. " Pp. 102-104, 154-156. 30 Pp. 73-76. 96-97, 101, 106. " Pp. 73, 75.
a Pp. 74, 104-108. M Pp. 74, 96-97. " P. 96.
10
6. Despite the inadequate training, confusing in-
formation and problems with instrumentation, one
operator did diagnose the stuck-open PORV soon
after he arrived at about 6 a.m. 35 He then di-
rected that the block valve for the PORV be closed,
thereby stopping the leakage. 36 In addition, within
hours after the core was uncovered, at least three
utility personnel correctly diagnosed that condi-
tion. 37 One of them was a member of the utility's
emergency command team. 35 He stated that it had
been generally recognized that the core may have
been uncovered for an extended period after 7
a.m. 39 Yet statements by other senior managers on
the utility's emergency command team suggest that
they never recognized that the core was uncovered
on the first day of the accident 40
B. SEVERITY
1. Three Mile Island was the most severe acci-
dent at a commercial nuclear power plant in the
United States.
2. The severity of a nuclear accident is measur-
able in term? of duration, extent of damage, re-
leases of radiation, near- and long-term adverse
health effects on both the public and workers, and
hazards of cleanup.
3. The accident at Three Mile Island was of
prolonged duration, resulted in severe damage to
the core, and left the Unit 2 facility highly con-
taminated by radioactivity. The cleanup task is
still in its early stages and. as described below, is
unprecedented in scope.
4. Three Mile Island is not the first serious acci-
dent at a nuclear reactor here or abroad. For
example, in 1957 there was an accident at the
Windscale plutonium-production reactor in Great
Britain that involved offsite releases 1.000 times
greater than those at TMI." Tn 1961, three workers
were killed in an accident at SL-1, a small research
reactor in Idaho. 42 An accident at the Enrico
Fermi power reactor in Michigan in 1966 resulted
in a partial core melt. 43
5. The Special Investigation reviewed some
available data and the findings of other investiga-
tions regarding radiation releases. It found no
persuasive evidence that releases during the acci-
dent resulted in adverse near-term physical health
effects or will result in statistically significant
adverse loner-term physical health effects. 44
The pending House-Senate conference report on
the FY 1980 XRC Authorization Bill contains an
amendment by this Committee directing the XRC
and EPA to conduct a feasibility study on acquir-
ing additional information for plant workers on
the incidence of any adverse long-term physical
health effects from the TMI accident. The 'State
of Pennsylvania is also attempting to acquire ad-
ditional information for the general population
bearing on the incidence of any adverse long-term
physical health effects from the accident.
The absence of evidence of major releases is
supported by conclusions of the Food and Drug
Administration based on its checking of photo-
graphic film in stores and facilities near the plant
for fogging caused by radiation. 45
Offsite radiation monitoring was both disor-
ganized and insufficient during the early hours of
the accident, making determination of actual re-
leases difficult. 4 ' A high percentage of the portable
radiation survey instruments were inoperable ; the
offsite dosimeters in place before the accident could
register only total radiation exposure over time
and not hourly dose rates ; management of the util-
ity's health physics program was inadequate. 47
Evidence of the limited extent of offsite releases
was developed by extrapolating from releases
measured at the boundary of the plant site and by
backcalculating from measurements of later off-
site releases. 48
6. There have been accidental releases of radia-
tion since the accident, but the Investigation found
no persuasive evidence that releases since the ac-
cident resulted in adverse near-term physical
health effects or will result in statistically signifi-
cant adverse long-term physical health effects. 49
The pending House-Senate conference report on
the FY 1980 XRC Authorization Bill contains an
amendment by this Committee directing the NRC
and EPA to conduct a feasibility study on acquir-
ing additional information for plant and cleanup
personnel bearing on the incidence of any adverse
long-term physical health effects from these
releases.
7. An important issue in determining the sever-
ity of the accident is whether the core was uncov-
ered more than once on the first day of the acci-
dent. If it was. the risk to the public was greater
than realized at the time, and there was even
greater reason to consider protective action. 50
The President's Commission concluded that
there was a second uncovering of the core in the
afternoon of the first day, when the utility was
attempting to depressurize the reactor coolant sys-
tem. 51 However, the XRC Special Inquiry Group
and the industry's Xuclear Safety Analysis Cen-
"Pp. 108-109. "P. 100. "Pp. 113. 116. "P. 116. "P. 116. "Pp. 116-117, 124-127, 129-130.
" P. 224. "P. 221. " P. 225.
" See. for example, ''Population Dose and Health Impact of the Accident at the Three Mile Island Xnclear Station."
Preliminary Estimates for the Period March 28, 1979 through April 7. 1979. XRC. XT'REG-0558. May 1979. pp. 60-63.
** P. 45. " P. 44. " P. 44. P. 44. " Post-accident radiation releases have been less than those during
the accident itself. See fn. 44. " For details, see p. 17 of "Findings and Conclusions." " P. 141.
11
ter 52 have interpreted the evidence to show there
was no second uncovering. 53
The Subcommittee believes that the available
evidence is insufficient to permit a final conclusion
on whether the core was uncovered a second time.
Further evidence to help resolve this issue may be-
come available at such time as the core is removed.
C. EMERGENCY PLANNING
1. Effective emergency preparedness requires
the assumption that serious accidents can happen
and that adequate plans need to be made in ad-
vance to deal with them. Such plans should be
based on a realistic consideration of the range of
potential accidents and must ensure that the re-
sources and procedures necessary for dealing with
such realistic contingencies will be readily avail-
able.
2. Prior to the accident at Three Mile Island,
emergency response planning was based on the
assumption that certain types of accidents those
involving disruption of the core (designated as
Class 9 accidents by the NRC) were so unlikely
that they did not need to be covered by emergency
plans. 54 Emergency planning was based on acci-
dents considered most likely ones of short dura-
tion that did not involve disruption of the core. 55
Further, the focus was on accidents involving the
failure of a single plant component, rather than on
multiple failures (two or more components) such
as occurred at TMI. 56
3. Prior to 1975, the NRC did not anticipate
playing an active emergency response role during
an accident. The agency saw its role as simply
monitoring the progress of an accident, using in-
formation provided by the utility. The NRC as-
sumed that an accident would be over before the
agency had a chance to get actively involved. 57
However, as a result of the duration of the fire
at the Browns Ferry nuclear power plant in 1975
and the agency's inadequate response the NRC
reevaluated its role. 58 A consultant's study con-
cluded that the NRC might need to take an active
role in managing an accident, and that a prereq-
uisite for that role would be the capacity to obtain
information, independent of the utility, about
plant conditions. The consultant recommended
that the NRC install an independent remote system
for monitoring plant conditions, tied directly to
NRC headquarters. 59
At the time of the TMI accident, the NRC had
identified an active managerial role as desirable
in the long-term, but had not installed the com-
munications system fundamental to fulfilling such
a role. 60
4. The accident at Three Mile Island was unlike
anything anticipated by the utility, the NRC or
the State. 61 None was prepared for an accident of
this nature and duration. 62 Events showed that the
utility, the NRC and the State did not readily have
available the resources essential for a proper re-
sponse. 63 According to the statements and testi-
mony of the participants in the crisis, a funda-
mental reason for their lack of preparedness was
conceptual : unduly narrow assumptions had been
made as to the kinds of accidents to be antici-
pated. 64
Thus:
Neither the utility nor the State's emer-
gency plans contained procedures provid-
ing for a continuous update of operational
data or of changing conditions in the status
of the reactor. 65
The NRC had conducted no drills of more
than a few hours duration at its Incident
Response Center. 66
The NRC's communications system pro-
vided for only one line to the Regional Of-
fice and did not cover direct contact be-
tween NRC headquarters and the Unit 2
control room; such contact was not estab-
lished until about 4:30 p.m. on March 28,
twelve and a half hours after the accident
began. 67
The NRC's regional emergency response
plan allowed up to six hours after notifica-
tion for the NRC to get its inspectors onsite,
a further indication that the agency was un-
prepared to take an active role onsite in a
timely fashion. 68
The State did not have enough technically
qualified staff assigned to its emergency re-
sponse organization. 69
5. There also were severe deficiencies in the or-
ganization and management of emergency re-
sponse planning within and among the three
organizations. These deficiencies went beyond
problems caused by the unduly narrow assump-
tions as to the kinds of accidents to be anticipated.
Thus :
The NRC headquarters and regional offices
had produced several incomplete and in-
82 Pp. 142-143. " Pp. 142-143. " Pp. 73, 75. " Pp. 83-84. M Pp. 73-75. " Pp. 83-84. " P. 82. * Pp.
82-83. "P. 83. "Pp. 83-84. "Pp. 73-74, 83-84. ^ Pp. 74, 120-121, 130-131, 160. M Pp. 73-74, 83-84.
" Pp. 134-135, 160. M Pp. 83-84. " Pp. 120-121, 131, 137. M Pp. 130-131. " P. 135.
12
compatible plans defining emergency re-
sponse. 70
The Commissioners never met as a body the
first day."
XRC emergency response plans were vague
about the Commissioners' role in an accident. The
Commissioners were to make policy as needed, but
that role was not defined with any specificity. This
was particularly true with respect to directing the
utility's response and to considering the need for
evacuation or other protective action. 72
There was no pre-planned coordination be-
tween the Commissioners in their offices in Wash-
ington, D.C. and the XRC emergency response cen-
ter in Bethesda. Md. on the day of the accident, A
system for briefing the Commissioners evolved on
an ad hoc basis over the day. 73
The utility's response was inadequate, par-
ticularly with respect to management and com-
munications. Statements by members of the
utility's emergency command team indicate that
many decisions during the first day were based
upon incomplete information because they failed
to share what they knew or believed about plant
conditions. 74
The Met Ed Emergency Plan provided no
guidance about how to assess the condition of the
plant during an accident. Further, it provided no
system for internal plant communications; it
merely delegated the responsibility for developing
internal communications procedures to the Emer-
gency Director. 75
There was no procedure in the Emergency
Plan for participation by the reactor-vendor, the
XRC or the architect-engineer in assessing plant
conditions. 78
The State had two organizations with three
emergency response plans covering accidents at
nuclear power plants, each of which differed in
significant respects and none of which conformed
to the utility's plan. 77
There was a lack of coordination among the
utility, the XRC and the State in their emergency
planning. 78
Although the XRC had provisions for Fed-
eral inter-ajjencv review of plans submitted volun-
tarilv by the States for XRC concurrence, the
XRC had not concurred in any of Pennsylvania's
plans. 79
The XRC, the utility and the State encoun-
tered severe communications difficulties involving
both the means of transmission and the quality of
information transmitted. This is further evidence
of insufficient joint emergency response planning. 80
6. Under the Atomic Energy Act, the NEC has
overall responsibility for the nealth and safety of
the public with respect to the operation of nu-
clear power plants. At the time of the accident,
however, there was no NRC requirement mandat-
ing that a State have an adequate emergency plan
prior to XRC licensing of a facility; nor a re-
quirement that the utility's plan be consistent with
the State's plan. 81
D. RESPONSES TO THE ACCIDENT
1. GENERAL
a. The responses of the utility, the XRC and the
State to the accident were inadequate.
b. Utility personnel, for the underlying reasons
discussed in I.A.2 above, proved unable to diag-
nose the accident correctly in time to prevent a
serious situation. 82 They took incorrect actions, ag-
gravating what began as a minor problem. 83 The
utility did not communicate effectively within its
organization or with the State and the NRC, par-
ticularly with regard to the possible need for evac-
uation or other protective action. 84
The utility's outside communications were poor,
leading Congressman Morris K. Udall to raise
questions as to "Why on March 28 ... [govern-
ment] officials and the public were denied impor-
tant information" about plant conditions. 85 The
XRC is still investigating this matter. The evi-
dence reviewed by the Special Investigation does
not confirm any intentional concealment of infor-
mation by the utility on the first day of the
accident. 86
c. The XRC was unprepared for an accident of
the duration and severity of that at TMI. It was
unable, during the first day, to contribute effec-
tively to either the diagnosis of the accident or to
developing strategies for achieving stability at the
plant. 87 It, too, was handicapped by highly defi-
cient internal and external communications. 88 Fi-
nally, at no point during the first day did the XRC
70 For example, in the emergency response procedures drawn up by NRC headquarters, a role was not defined for
the regional office in the integrated agency -wide response (see p. 80) ; the regional plan envisioned the regional office
as the lead unit within the NRC and did not state how its response would relate to that of headquarters (see p. SO).
The XRC's manual assigned the Office of Nuclear Reactor Regulation (NRR) specific responsibilities and called for
the Office of Inspection and Enforcement (I&E) to define the implementing procedures for the entire agency's emergency
response. (See p. 81.) Yet, the implementing procedures prepared by I&E did not include a role for NRR. During the
accident. NRR and I&E had essentially separate emergency response teams (see pp. 157-158).
"Pp. 119, 131. n Pp. 81-82. "Pp. 79, 131. "Pp. 116-117, 124-127, 138. 141. "P. 160. "P. 160.
" Pp. 84-85. Pp. 84-86. P. 84. " Pp. HOff. n P. 84. Pp. 94-109. " Pp. 94-109. " Pp. 113ff.
" Pp. 113-117, 138-141. M Pp. 138-141. " Pp. 145, 147-149. * Pp. 119-120, 127-128, 130-132, 137-138, 143, 145.
13
give serious consideration to recommending pro-
tective action. 89
d. The State did not actively solicit the infor-
mation it needed to make independent judgments
about plant conditions. 90 Rather, it simply relied
on incomplete and often inaccurate information
supplied oy the utility. As a result, the State,
which has primary responsibility for ordering pro-
tective action, did not appreciate the serious need
to consider such action. 91
e. A review of all the responses discloses three
basic deficiencies :
Pre-accident emergency response planning
was inadequate.
Transmittal of information was badly mis-
handled.
Failure to perceive the need for serious con-
sideration of protective action was a major over-
sight.
2. THE UTILITY'S RESPONSE
a. During most of the first day of the accident,
plant operators and managers, 92 according to
their statements, failed to diagnose the plant's con-
dition in particular, the loss of core coolant dur-
ing the initial hours, and the subsequent uncover-
ing of, and severe damage to, the core. 93 Control
room personnel did not systematically bring to-
gether, review and track plant conditions. Such
actions would have helped them in diagnosing the
status of the reactor. 94 In some instances, they said
they discounted plant behavior and indicators that
suggested the core had been imcovered and dam-
aged. 95 Their statements indicate that to the extent
they discussed key symptoms or events, they did so
without analyzing causes or possible conse-
quences. 96
These failures contributed to actions by control
room personnel that led to a worsening of the ac-
cident and that contributed to its duration.
b. The actions of the plant operators and
managers must be analyzed in the context of
deficiencies in training, 97 control room design, 98
instrumentation and equipment, 99 plant design, 100
and emergency procedures, 101 as well as the stress
and confusion produced by the crisis. 102
c. One reason control room personnel failed to
diagnose plant conditions correctly was that key
readings of temperatures in the core were rejected.
Instruments from which they had been taken the
incore thermocouples, which provide core coolant
temperatures were thought to be unreliable.
Some thermocouples had failed, but others re-
mained operational and were in fact giving the
only direct and reliable readings of core tempera-
tures. 103 The director of the utility's emergency
command team said he was advised by the lead in-
strumentation engineer that an initial set of five
thermocouple readings indicated all the thermo-
couples should be considered unreliable 104 advice
which proved to be incorrect. 105 The team did not
receive a subsequent set of readings taken from all
52 thermocouples, although such information was
available. 106
Had plant operators and managers considered
the thermocouples reliable, they would have had a
clear signal that the core was or had been un-
covered. The thermocouples also would have been
useful in tracking the success of attempts to return
the plant to a stable condition.
Another important instrument, the movable in-
core detector, also could have been used prior to
severe core damage to help determine whether the
core was covered and whether operating strategies
were effective. Some utility personnel said they
considered it to be a device for use by the reactor-
vendor. 107 The instrument was not used the first
day.
d. According to accounts by control room per-
sonnel, there were other instances in which they
missed, misinterpreted or discounted critical in-
formation 10S and in which critical information
was not communicated to and among key person-
nel, resulting in fragmentaion of information that
impeded an effective response. 109 An example was
the response to the hydrogen burn in the contain-
ment at 1 :50 p.m. This burn was a symptom of
uncovering of, and damage to, the core. There
were several indicators of the burn, including 110
An unusually high reading of containment
pressure the "pressure spike" which ap-
peared on a strip chart in the control room ;
Automatic start-up of cooling sprays in the
containment ;
Automatic isolation of the containment, in
response to the high pressure ;
Automatic actuation of the high pressure
injection system (a part of the Emergency
Core Cooling System) ; and
An unusual noise heard in the control room.
" Pp. 119-120, 132-134, 146-147. * Pp. 121, 135-136. " Pp. 121, 135-136.
K Four control room personnel were on duty when the accident began and were responsible for the utility's imme-
diate response. They were joined by a supervisor and two engineers within minutes, and later by other engineering and
supervisory personnel. At 8 a.m. the utility established an emergency command team, which included a representative
of the reactor-vendor. These control room personnel made decisions for the first 12 hours. ( See p. 112. )
93 Pp 102-104, 120, 124-127, 129-130. M Pp. 99-109. 113-114, 141-143. "Pp. 113-114. 116-117, 124-127, 138-142.
80 Pp. 113-114, 124-127, 138-142. " Pp. 73-76. * Pp. 56-64. " Pp. 65-66. 69-71, 94, 96, 99-101, 103, 104, 155.
100 Pp. 96, 99-100, 125-126. *" P. 58. 102 Pp. 97-101, 105, 109, 117, 124-127, 138-140. m Pp. 116-117.
104 Pp. 113-114, 116-117. 1<e Pp. 116-117. 1M P. 114. "" Pp. 74, 112. "" Pp. 117-118, 124-127. 1M Pp. 102-
104, 124-127, 143-144, 151. " Pp. 138-141.
14
Some control room personnel said they were
unaware of any of the symptoms. 111 Those who
said they were aware of most of the symptoms
suggested that they had focused only on the pres-
sure spike on the strip chart and that they had
discounted it as an electrical malfunction. 112 Only
one person said he concluded there had been a
hydrogen burn. 113 Utility personnel maintained
that they did not conclude the symptoms repre-
sented cumulative evidence of a core that had
been uncovered and damaged. 114
e. As noted, the failure of the utility to transmit
accurate information on plant conditions during
the first day. particularly regarding the hydro-
gen burn, has led to questions about whether the
XRC. the State, and the public were denied im-
portant information by the utility.
The weight of the evidence does not support in-
tentional concealment of information by the util-
ity on the first clay of the accident. There are con-
flicting statements as to whether the director of
the utilitv's emergency command team was made
aware of major evidence of uncovering of. and
severe damage to. the core. 115 On balance, however,
the evidence indicates that neither he nor other
utility personnel deliberately withheld this infor-
mation. In fact, the actions of these personnel
during the first day of the accident indicate that,
for all the underlying reasons discussed in I.A.2
and I.D.2.b above, they did not know or fully
understand the information available to them.
They were unprepared for. and unable to respond
effectively to. the emergency.
3. THE NRC'S RESPONSE
a. The XRC's response during the critical hours
of the first day was inadequate. The XRC did not
contribute effectively to the utility's effort to diag-
nose conditions at the plant, nor did it provide
guidance to the utility. 116 Though responsible for
public health and safety, the XRC did not ade-
quately consider evacuation or other protective
action, nor did it advise the State in this area. 117
This was so even though on at least two occasions,
key members of its emergency response organiza-
tion expressed their belief that the core was or
had been uncovered. 115 a situation clearly necessi-
tating consideration of protective action. 119
b. The XRC confined itself to monitoring events
at the plant, relyine on the utility for data on
plant conditions. 120 For most of the first day. the
XRC was unable to carry out even this limited
role effectively because it lacked accurate data. 111
Information was frequently unavailable, incom-
plete or garbled in transmission to both regional
and headquarters staff and between the regional
staff and headquarters, 112
c. Xo representative of the XRC was in the con-
trol room of the crippled reactor until 11 :30 a.m..
over seven hours after the accident began. 123 Even
then, the XRC onsite team failed to obtain, assess
and transmit in a timely fashion information
on key aspects of plant conditions, including
superheated steam in the hotlegs, 124 the utility's
inability to depressurize to the point at which the
decay heat removal system could be used. 125 and
the symptoms of the nydrogen burn. 12 ' Both the
onsite team and the regional office also failed to
obtain in a timely fashion accurate information
in response to specific requests from XRC head-
quarters, including those for incore temperatures
and for strategies being pursued by the utility. 117
d. The activities of the onsite and offsite teams
were poorly managed. 128
The XRC. both on- and offsite. did not have a
systematic method for asking pertinent questions
of the utility or for following up on issues raised,
especially about whether the core was covered. 129
At XRC headquarters, there was little, if any,
coordination among the components of the
agency's Incident Response Center. 130 XRC head-
quarters personnel who received important infor-
mation did not systematically transmit it to de-
cisionmakers at the Center or to the Commission-
ers. 131 As with the utility, there was no effective
means for assuring that each of the responsible
decisionmakers received and understood signifi-
cant information, such as indications of super-
heated steam in the reactor and its implications. 132
The XRC Commissioners exercised virtually no
oversight of senior staff at the Response Center
on the first day. 133 Their assumption had been that
any possible accident would not be of sufficient
duration to permit their active involvement, and
they were not prepared for that role. 1 "
e. The Acting Chairman, a member of the Com-
mission since its inception, was unfamiliar with the
XRC's emergency response organization and its
responsibilities. 135 He served as Acting Chairman
the first day because the Chairman, the most tech-
nicallv qualified member of the Commission. 136
was absent for personal reasons. 137 The remaining
members of the Commission either were unaware
'"P. 140. m P. 141. "P. 140. U4 Pp. 138-141. m Pp. 138-141. "P. 145. "'Pp. 133-135.
Pp. 119-120, 145-148. "* Pp. 85-86. 133-135. "Pp. 137-138, 145-1 46. "Pp. 119ff. m Pp. 119. 127-128.
131-132.137-138.143-144. 1=> Pp. 130-131. " Pp. 145-148. m Pp. 141-144. 147. M Pp. 138-141. "Pp. 128.
137. 145. 159. '* Pp. 137. 140. 145-147. 160. Pp. 137. 140. 145-148. 160. "* Pp. 146. 157-159. m Pp. 145-148.
" Pp. 145-148. m Pp. 150-151. " P. 151. m Pp. 133-135. 146-147.
"Nuclear Regulatory Commiss'on Special Inquiry Group, Three Mile Island: A Report to the Commissioners and
to the Public. Volume II. part 3. p. 933.
m P. 119.
15
of available information on the plant's condition
or did not understand its significance. 138 One Com-
missioner was told by an NRC staff member at
9 a.m. that the core probably had been uncovered
and that the state of the reactor was uncertain.
Yet neither the Commissioner nor the staff mem-
ber raised the issue of protective action. 139
f . Information communicated by the NRC to the
White House and other Federal agencies during
the first day was incorrect and misleading. Dur-
ing the afternoon, senior staff in the Center be-
lieved that plant conditions were unstable, and
they were concerned that the core was uncovered. 140
Yet, during this same period, the Response Center
informed the White House Situation Room and
the Department of Health, Education, and Wel-
fare that the utility was having no trouble keeping
the core covered. 141
This serious error, which has not been satisfac-
torily explained, 142 served to preclude the Presi-
dent and Federal officials from considering the
need to mobilize Federal resources to assist the
State and the NRC on the first day.
4. STATE AND LOCAL RESPONSE
a. The State's response was inadequate because
of deficiencies in its plans, insufficient information,
fragmentation and lack of resources, and poor
management. 143 As a result, the State did not ap-
preciate the serious need to consider evacuation or
other protective action on the first day. 144
b. The State's emergency plans led it to rely on
the link between a State environmental agency
the Bureau of Radiological Protection (BRP)
and the utility for information about the plant. 145
This made the State dependent on the utility for
such information. 146 Furthermore, the BRP had
only one technically qualified person familiar with
plant operations, a nuclear engineer. He was fre-
quently called away to brief the State's emergency
management group, and thus was not available to
request and to analyze information from the
plant. 147
c. The State's response to the accident was man-
aged by a group of State officials that had been
organized that day and that had not been desig-
nated in any of its emergency response plans. This
group failed to communicate information on the
status of the reactor to either State or local
agencies. 148
d. The State is ultimately responsible for deter-
mining whether protective action is necessary and,
if so, for ordering and implementing it. 149 The
principal reason the State did not not perceive the
serious need to consider protective action on the
first day is that it did not receive accurate informa-
tion on the severity of the situation at the plant.
The utility did not provide the ongoing informa-
tion on plant conditions necessary to determine the
need for protective action, and the State did not
solicit it. 150 State officials saw their role as acquir-
ing data on actual radiation releases that they
deemed to be the determining factor as to whether
protective action was needed. 151
5. INFORMATION TRANSFER
a. During the first 12 hours of the accident, a
significant amount of information was mishandled,
as a review of the responses of the utility, the
NRC, and the State makes clear." 2 Accurate in-
formation on the following plant conditions was
lost at one or more points in the chain of com-
munications :
Lack of natural circulation ; 153
Superheated steam in the reactor; 154
Concern that the core was uncovered ; 155
The correct set point for decay heat removal
system ; 15e
Hotleg temperatures ; 157
Incore temperatures ; 158
Symptoms of the hydrogen burn ; 159 and
Strategies being pursued by the utility to
stabilize the reactor. 160
Information was lost within the utility, 161 be-
tween the utility and the State, 162 between the
utility and the NRC's regional office, 163 between
the utility and NRC headquarters, 164 between the
utility and NRC onsite representatives, 165 between
NRC onsite representatives and the regional
office, 166 between the regional office and head-
quarters, 167 between members of the headquarters
senior staff, 168 between senior staff and the Com-
missioners, 169 and between senior staff and other
Federal agencies. 170
b. As predicted in the consultant's study follow-
ing the Browns Ferry fire, one reason the NRC lost
information was that it had not established a com-
138 Pp. 131ff.
M Pp. 135-136.
189 Pp. 119-120. 140 Pp. 145-148.
145 Pp. 84, 1 22. 1W Pp. 84-86, 135-136.
151 Pp. 85-86, 135. 1K Pp. 113ff.
'P. 149. ""P. 149. '"Pp. 84-85, 121-123, 135-136.
"' Pp. 135-136. '" Pp. 121-123. "' Pp. 135-136, 159.
' Pp 135, 159 1M Pp. 85-86, 135. "* Pp. 113ff. ira Pp. 120. 131, 143-144. I54 Pp. 114, 124-127, 142, 145-148.
'Pp 114 116 120, 126. 128-129. 132, 145. IS6 P. 129. "'Pp. 119, 132. 137-138. '" Pp. 113-114, 116-117, 137.
145. '"Pp. 138-141. ""Pp. 127-128, 137. lel Pp. 113-114, 116-118. 124-125. 127, 138-140. 142. "" Pp. 121,
135, 159. 16S Pp. 118, 120, 127, 130-132, 145. IM Pp. 143, 145. '" Pp. 129, 138-140. "* Pp. 137, 140. '" Pp. 119,
131-132, 137, 145. 1M Pp. 132-133, 146. 1<l9 Pp. 119, 133-134. " Pp. 120, 149.
16
munications system independent of the licensee. 1 "
The XRC did not heed prior recommendations for
direct "hotlines" to operating reactors and for
direct transmission of plant data offsite. 172
c. The accident demonstrated, however, that
adequate communications technology' will not of
itself ensure proper transmission of information.
In the afternoon, senior XRC officials were ques-
tioning whether the core was uncovered. 173 When
they finally obtained a direct link to the control
room at Unit 2, a senior XRC staff member com-
municated his concern only to an XRC inspector
in the control room. He did not pursue the ques-
tion directly with the utility personnel. Xor did
he ask the inspector to pursue the matter with the
utility. 174 despite a suggestion from the Acting
XRC' Chairman that he do so. 175
d. Implicit in the consultant's recommendation
that the XRC improve its communications system
was a recognition of the limits on human perform-
ance imposed by stressful conditions. Statements
by utility operators and managers suggest that
their confusion and anxiety under stressful condi-
tions was an important factor contributing to the
loss of information and failure in commu-
nications. 1715
e. Control room personnel were uncertain of the
status of the reactor for a prolonged period on
the first dav of the accident. This uncertainty was
itself a "plant condition'' that should have been
clearly communicated to the State and the XRC
and used as a factor in determining the need for
protective action.
6. PROTECTIVE ACTION
a. On the rlav of the accident, the emergency
plans of the utility. 177 the XRC 178 and the State, 179
as well as the Environmental Protection Agency's
(EPA) Manual of Protective Action Guides, 180
were inadequate or incomplete regarding either:
1) factors to be weighed in projecting dose
rates so that adequate consideration could be
given to the need for evacuation or other pro-
tective action:
2) information that a utility was required
to communicate to the XRC. the State and the
public during the accident.
b. The actions of the utility, the XRC and the
State on the first day indicate that, in determining
the need for protective action, they relied too
heavily on the radiation dose levels specified in the
EPA Protective Action Guides. 181 Xone ade-
quatelv focused on plant conditions including
uncertainty as key factors in projecting doses for
determining whether action was needed to protect
the surrounding community. 182
c. The EPA has legal responsibility for pro-
viding guidelines to the States and utilities on
protective action. Therefore, it must shoulder some
responsibility for inadequacies in protective ac-
tion decisionmaking during the accident.
The 1975 version of the EPA Manual makes one
ambiguous reference to plant conditions as a fac-
tor to be used in projecting doses, without specify-
ing their importance. 183 A January 1979 revision
of the Manual (and a further revision made fol-
lowing the accident) make plant conditions a cru-
cial element in formulating projected doses, but
still do not define the term "plant conditions." 184
The various versions of the Manual also fail to
specify any role for the XRC and do not provide
guidance to the States and utilities as to how de-
cisions on protective action are to be reached in
the event there is uncertainty as to what is oc-
curring at the plant. 185
d. Although the EPA guidelines were ambigu-
ous and incomplete, it should be stressed that even
in the absence of any guidelines, the utility, the
XRC and the State nevertheless should have given
great weight to plant conditions, particularly the
uncertainty about uncovering of the core, as being
a determining factor in considering the need for
evacuation or other protective action.
The utility was remiss in not clearly communi-
cating its uncertainty on the morning of the first
day to the XRC and the State. At the same time,
the XRC and the State were remiss in failing to
pursue effectively with the utility the issue of
plant conditions, including most particularly its
uncertainty about whether the core was uncovered.
Statements by members of the utility's emer-
gency command team indicate that some of them
were uncertain about whether the core was un-
covered at 8 :30 a.m. on the first day of the acci-
dent. It is unclear from statements by the team
leader responsible for recommending protective
action as to whether he was among those who were
uncertain during this period. Two weeks after the
accident, he said: "Based on the instruments we
had, we didn't know if the core was covered." 18S
Subsequently, he said: "... I didn't believe the
core was uncovered, but I listened to people in my
group looking for double assurance." 18T
e. If the utility official responsible for recom-
mending protective action had properly under-
stood his role, as defined by the EPA Manual
and the utility's emergencv plan, and if he had been
substantiallv uncertain, based upon plant condi-
tions at 8:30 a.m., about whether the core was
171 Pp. 82-83. m Pp. 82-83. m Pp. 145-148. " 4 P. 147.
79. 136. 160. " Pp. 79-82. ' Pp. 84-85. ** Pp. 134-135.
86. '" Pp. 85-86. '* Pp. 85-86. '"P. 114. " P. 129.
P. 146.
' Pp. 133-136.
'Pp. 126-130. 132. 137-139. '"Pp.
m Pp. 85-86. 134-135. 1B Pp. 85-
17
uncovered, he then should have advised State offi-
cials that the condition of the plant at that time
warranted consideration of a possible precaution-
ary evacuation of the population within a close
proximity of the plant.
E. PRIOR OPERATING EXPERIENCE
1. AT OTHER NUCLEAR PLANTS
a. Three Mile Island was not the first nuclear
facility to experience the conditions that occurred
in the early stages of the March 28, 1979 accident.
Important information had been available to the
reactor-designer of TMI and to the NRC on minor
accidents at two other plants Oconee in South
Carolina and Davis-Besse in Ohio 188 that were
similar to the beginning of the TMI accident.
b. Both the reactor-vendor, Babcock & Wilcox
(B&W), and the NRC had programs for evaluat-
ing and acting upon individual problems occur-
ring at nuclear power plants. However, the
responses of the reactor-vendor and the NRC to
these similar accidents suggest that neither had
procedures to assure an effective systematic review
and analysis of potentially recurring problems. 189
For these reasons, TMI control room personnel
did not have the benefit of analysis and guidance,
based on similar accidents, that would have helped
them in diagnosing and responding correctly to
the earlv events of the accident on March 28. 190
The deficiencies in industry and NRC programs
for evaluating and acting on operating experience
at nuclear power plants were among the most
important inadequacies in the nuclear safety pro-
gram brought to light by the accident.
2. AT TMI-2
a. Plant behavior during two incidents in the
early testing and operating history of TMI Unit
2 was similar to plant behavior that TMI control
room personnel failed to understand during the
earlv hours of March 28. 191
The first occurred in 1977 during "hot functional
testing" of the plant. 192 Steam collected in the
hotlegs of the reactor's primary coolant system,
causing the water level in the pressurizer to in-
crease as pressure in the system decreased. Details
of this earlier event apparently had not been com-
municated to the operators on duty during the
early hours of the March 28, 1979 accident. They
neither recognized nor understood similar condi-
tions on the day of the March 28, 1979 accident. 193
The second incident occurred on March 29, 1978.
A temporary loss of power to an electrical con-
trol system caused the pilot-operated relief valve
(PORV) to open, allowing water to escape from
the primary coolant system. The operators on duty
did not know the valve was open because there
was no indicator of its position in the control
room. 194
Subsequently, the utility installed a command-
type indicator that would show whether an elec-
trical signal was being sent to open the valve, but
it did not show the valve's actual position. The
operators stated that this type of indicator was
less desirable than the one they had requested to
show actual position directly. 195
During the March 28. 1979 accident, plant per-
sonnel did not realize for more than two hours
that the PORV was stuck open. One reason was
that the operators took the absence of the light in-
dicating a command to open the valve as evidence
that the PORV was closed. 196
b. Two other aspects of the operating experi-
ence of Unit 2 of the Three Mile Island plant con-
tributed to the failure of plant operators and
managers to diagnose the early symptoms of the
accident correctly.
First, it was known by plant personnel that one
or more of the valves on top of the pressurizer
had been leaking coolant water for more than six
months. Because of the leakage, the temperature
readings for the discharge line leading from the
valve were abnormally high during normal oper-
ations. 197 Statements of control room personnel
indicated they had become accustomed to these ele-
vated temperature readings. 198
During the early hours of the accident on
March 28, temperature readings in the line rose
even higher after the PORV opened. When the
valve stuck open, they remained at a higher tem-
perature than the operators were accustomed to.
Nevertheless, control room personnel were misled
by the anticipated "normal" high readings and
by their knowledge that the PORV had lifted
briefly. They, therefore, failed to recognize that
the readings during the accident were indicating
'* Pp. 76-77.
* In assessing a fine of $100,000 against the reactor-vendor, the XRC ". . . concluded that B&W did not have an
effective system for collection, review and evaluation, and reporting of important safety information." (Letter from
Victor Stello, Jr., Nuclear Regulatory Commission, to J. H. MacMillan, Babcock & Wilcox. re : "Notice of Noncompli-
ance," April 10, 1980.) In response, B&W denied the charges, but paid the fine. (Letter from IX K Gilbert. Babcock &
Wilcox. to Victor Stello. Jr., Nuclear Regulatory Commission, re : "Notice of Noncompliance," May 20. 1980.)
'* Pp. 77-78. 07. 101. m Pp. 65-66. m P. 65. '" P. 97. "" P. 66. '" P. 66. '" Pp. 94, 155.
'" Usually, high temperatures indicate the valve has opened. Sustained high temperatures indicate that it is stuck
open or that there is a slow leak.
"" Pp. 108, 156-157.
18
a continuing and significant loss of coolant
through the stuck-open PORV. 1 * 9
Had the utility closed the block valve, as re-
quired by the plant's Technical Specifications, the
loss-of-coolant accident through the stuck-open
PORV would not have occurred. Had the utility
corrected the PORV leakage, the operators would
have been in a better position to determine that
the elevated temperature readings indicated a Loss
of Coolant Accident.
Second, although high pressure injection
(HPI) is designed to actuate under loss-of-
coolant conditions, there was a history at TMI-2
of actuation in response to relatively routine
problems in the secondary system. 200 Plant per-
sonnel had become accustomed to initiation of
HPI in response to these less significant incidents.
On March 28 they throttled HPI before de-
termining whether there was a loss-of-coolant
accident. 201
II. RECOVERY
A. GENERAL
1. The recovery process at Three Mile Island
will take place in two stages: (a) cleanup of the
radioactive debris from the accident and (b) final
disposition of the plant- either refurbishing it as
a nuclear or coal facility or permanently decom-
missioning it. 202 The Special Investigation ad-
dressed principally the cleanup, since discussion
of the future of the facility can be only specula-
tive at this time.
2. Cleanup is a large, potentially hazardous and
technically difficult task. I^arse quantities of
radioactive gases and water within the contain-
ment must be removed, as must damaged fuel and
other contaminated material in the reactor vessel.
All the radioactive waste must be disposed of.
3. Cleanup is not. however, simply a technical
task. It involves many other factors. Financial,
social and legal issues were addressed by the Spe-
cial Investigation. All have a bearing on cleanup
decisions.
4. The damaged plant at Three Mile Island
must be decontaminated. 77<??r and irhen. however,
are still unresolved.
The timing of the various steps of cleanup
poses a dilemma. It is desirable to follow de-
liberate procedures providing for review of al-
ternatives, for orderly decisionmaking and for
public participation. 103 Yet as time passes, there
is an increasing chance of accidental releases of
radioactivity to the environment 204 and perhaps
even of renewed fissioning (recriticality) of the
damaged reactor core. 205 At present, the plant's
condition is not fully known. Further deteriora-
tion can be assumed. Damaged and unmaintained
equipment may fail, and there is the potential for
human error.*"*
Both the surrounding community and, most im-
mediately, the workers involved in cleanup are at
risk. 207 These workers will continue to be exposed
to radiation as long as the plant remains contami-
nated. 206 The hazard to them of accidental overex-
posure will be present as long as areas of high
radiation are widespread. 209 As noted, the pending
House-Senate conference report on the FY 1980
XRC Authorization Bill contains an amendment
by this Committee directing the XRC and EPA
to conduct a feasibility study on acquiring addi-
tional information for plant and cleanup person-
nel bearing on the incidence of any adverse
long-term physical health effects from these
exposures.
B. TECHNICAL ISSUES
1. The technical aspects of the cleanup at Three
Mile Island present unprecedented challenges. For
example, certain problems require the develop-
ment of new equipment and techniques. 210 The
most difficult of these is removing and disposing
of the damaged core. 211
However, measurements of samples from inside
the containment building indicate that some an-
ticipated technical difficulties in the cleanup, such
as the amount of radioactive cesium on the inner
walls of the containment building, will not be
as extensive as feared, thus simplifying some as-
pects of the overall task. 212 Recently, on the other
hand, the unsuccessful first attempt to enter the
containment (the access door was stuck) raises
the question of whether unforeseen problems, in-
cluding corrosion, will make decontamination
more difficult.
Similar problems have been solved in cleanups
of other nuclear accidents in this country and
abroad. 213 But there are differences. First, the
cleanup problems at TMI are of a larger scale than
ever experienced in the commercial nuclear power
program. 214 Moreover, at TMI. unlike with those
"Pp. 71. 10v "P. 72. "Pp. 72. 96. * Pp. 188. 190. " Pp. 163. 205. 206. "* Pp. 164. 166.
105 An XRC report assessed the ways in which recriticality might occur and what the consequences would he. The
report found that the most likely radiological consequence of recriticality would be increased dose rates inside the con-
tainment and that offsite consequences probably would be non-existent. Hence, the XRC found that the risk is to the
workers. See also pp. 165. 166.
"* If many years pass before appreciable cleanup progress is made, the chance of accidents, including recriticality.
accumulates, since there will be rouphly 1.800 workers onsite. all capable of human error that could trigger such
accidents. S>e also pp. 164. ivT,.
*" Pp. 166-167. 175-177. " P. 176. " P. 163. *" Pp. 168. 187-188. nl Pp. 168, 175. 187. M P. 186.
111 Pp. 219-226. * Pp. 169. 221.
19
prior accidents, private rather than governmental
entities bear primary responsibility for accom-
plishing the cleanup task. 215 Finally, the TMI
cleanup is taking place within the context of re-
cent environmental review requirements, including
those for public hearings. 216
More than one year after the accident, many un-
certainties remain over the future course of clean-
up. As late as early June 1980, the utility had not
yet entered the containment in order to conduct
a more detailed evaluation. 217
C. FINANCIAL ISSUES
1. General Public Utilities Corporation (GPU)
and its three subsidiary utility companies face
financial problems as a result of the accident, as
evidenced by the sharply decreased value of GPU
common stock and by the downgraded bond
ratings given the utilities. 218
A major expense has been the purchase of re-
placement power. The three GPU subsidiary com-
panies Metropolitan Edison Company (Met Ed) .
Jersey Central Power & Light Company (Jersey
Central) and Pennsylvania Electric Company
(PENELEC) have had to purchase electric
power to replace the output previously provided
by the damaged Unit 2 and by Unit 1, which has
not been operating since the accident. 219 Replace-
ment power costs ranged from $20 to over $,35 mil-
lion per month during 1979. 220 The utilities lack in-
surance to cover this expense and have requested
and received considerable rate increases from the
public utility regulatory agencies in Pennsylvania
and New Jersey to help cover these costs. 221
GPU has estimated that costs of cleanup alone
will total at least $200 million. One manage-
ment consultant has said that the final figure could
be $500 million or more. 222 The utilities have $300
million in property damage insurance to offset
part or all of the cleanup cost. 223
The Pennsylvania and New Jersey public util-
ity regulatory agencies have removed the capital
and operating costs associated with Unit 2 from
the utilities' rate basis. 224 Thus, the utilities' cus-
tomers are not paying for cleanup costs. 225
2. To cover immediate expenses, the utilities
have borrowed substantial sums from a con-
sortium of banks. 226 GPU and bank officials as
well as officials from the Securities and Exchange
Commission all have stated that Met Ed's con-
tinued solvency may depend on favorable rulings
by the State public utility regulators. 227
Utility regulators in Pennsylvania and New
Jersey have acknowledged the importance of their
actions. 228 Thus far, they have indicated their in-
tention to provide the rate relief needed to help
preserve the financial viability of the three utili-
ties. 229 In its May 23, 1980 decision, the Pennsyl-
vania Public Utility Commission stated that it
was providing Met Ed an "adequate framework"
for financial recovery and that it was up to Met
Ed to convince bank creditors that it had "the will
and the ability to rehabilitate itself." 23
3. Given the financial situation, New Jersey
utility regulators are considering alternatives to
Jersey Central's existing operations. 231 For the
same reason, Pennsylvania utility regulators con-
sidered withdrawing Met Ed's certificate of pub-
lic convenience (its franchise to serve the
public) , 232 but in May 1980 decided that the public
welfare would not be well-served by modifying
or revoking it. 233 At the same time, they affirmed
their authority to reconsider the issue. 234
4. The financial future of the GPU companies
also will be affected by an ongoing NRC regula-
tory proceeding to determine whether TMT Unit 1
will be returned to service. 235 Its resolution will
affect the extent to which the utilities must con-
tinue to purchase and require rate relief to
cover replacement power and whether the capi-
tal and operating costs of Unit 1 may be returned
to the utilities' rate bases. 23 " Hearings are not
likely to begin before the fall of 1980, and no firm
date for a final decision has been set. 237
5. There have been few instances of bankruptcy
proceedings involving major electric utilities. The
GPU companies' financial condition, however, has
raised this possibility. 238
An SEC official stated that, as a practical mat-
ter, a utility providing electric power to the pub-
lic would not be closed down. 239 GPU stated that
Met Ed's bankruptcy would not be in the parent
company's best interest. 240 Banks similarly testi-
fied that bankruptcy was not in the lenders'
interest. 241
6. The three operating utilities and their cor-
porate parent, GPU, acknowledged their obliga-
tion to clean up the TMI site. 242 However, they
declined to make a firm, blanket commitment to
do so without regard to future circumstances, par-
ticularly bankruptcy of Met Ed. 243
m Pp. 168. 221. "'Pp. 163, 202-207. ! " P.
' Pp. 191, 193, 212-216. "'P. 190. K 'P. 191.
184.
'P. 190.
'Pp. 190-191, 194, 212.
'P. 190.
Ki In April and Mav 1980. the two state public utility commissions also removed the Unit 1 capital and operating costs
from the utilities' rate bases ; see p. 191, fn. 93, and pp. 212-216.
Ki Pp. 191, 212-213, 215.
193, 214. 2 "Pp. 215-216.
1 Pp. 191-192.
a Pp. 214-215.
'" Pp. 192-193.
:3a Pp. 214-215.
' Pp. 193. 214-216. m Pp. 193, 214-216. !M Pp.
M4 P. 215. *"Pp. 194, 212. 2M Pp. 194, 212.
237 Pp. 194, 212.
a 'Pp. 193. 194, 214, 216: see letter from Grant G. Guthrie, Division of Corporate Legislation. SEC, to Jonathan
Cottin, TMI Special Investigation Staff, November 2. 1979.
"* P. 194. P. 194. *' P. 194. M P. 195, fn. 105. J4> Pp. 194-195.
20
Bankruptcy of Met Ed could further complicate
cleanup. 244 Yet, the XRC only recently acknowl-
edged the need to prepare plans for this contin-
gency. 245 According to the testimony of XRC
officials, the agency has the necessary legal au-
thority to assume that responsibility. 246 However,
they also stated that the agency had manpower
resources only to manage cleanup activities, not to
man all of the equipment. 247
D. SOCIAL ISSUES
1. Concerns of the communities surrounding
TMI have complicated cleanup. Many residents
and local officials have expressed strong distrust
of the utility and the XRC and have questioned
the ability of those two organizations to manage
the cleanup and to be candid in discussing prob-
lems and risks. 248
Some of this distrust and anxiety is attributable
to events during the March 28 accident and to the
cleanup, including several reported accidental
radiological releases. 249 Another factor is the fail-
ure of the XRC and the utility to agree on a defini-
tive cleanup plan and the continuing attention
TMI has received from the media and survey-
takers. 250 In addition, there have been complaints
that the XRC and the utilities failed to notify lo-
cal officials of planned cleanup activities. 251
2. The XRC and the utility have been unsuc-
cessful in their efforts to increase public confidence.
In late 1979. XRC and utility officials began hold-
ing biweekly public meetings so that cleanup
plans could be discussed and residents could ask
questions. 252 In addition, once the XRC decided
to prepare an environmental impact statement,
the agency held public meetings to discuss the
scope of this document. 253 By early February
1980. the XRC also set up a permanent office in
Middletown to keep in closer touch with events at
the site and local concerns. 254
Yet in late February, an XRC task force on
cleanup concluded that segments of the commu-
nity continued to have strong feelings of fear and
anxiety. 255 The task force suggested that the
agency consider funding a citizen's advisory
group. 256
Early in April, some citizens became upset with
an XRC staff recommendation to vent the krypton
in the containment building. 257 Governor Richard
Thornburgh asked the Union of Concerned Scien-
tists, a group opposed to nuclear power, to review
the XRC staff recommendations, 258 reflecting an
attempt to find a technically competent third party
whose recommendations concerning cleanup would
be accepted by those who remained distrustful of
the XRC and the utility. 259 Following review of
a number of independent studies, including the
Union of Concerned Scientists' study, the Gov-
ernor decided to support the XRC if it should
decide to vent the krypton. 260
Opposition to cleanup is by no means unani-
mous, however. Some local citizens are critical of
the vocal opponents of cleanup and are anxious
to proceed promptly with the cleanup steps pro-
posed thus far so that the process can be expe-
ditiously completed. 261
E. LEGAL AND REGULATORY ISSUES
1. The legal and regulatory procedures appli-
cable to the cleanup effort at TMI are intended
to assure reasonable decisions through considera-
tion of all relevant factors and viewpoints. How-
ever, the process results in cleanup proceeding at
a deliberate pace. This creates a dilemma. The
longer it takes to remove the radioactivity from
inside the plant, the more likely it is that further
accidental releases of radioactivity will occur be-
fore workers can repair or remove deteriorating
equipment. Although the XRC has limited au-
thority to act before the completion of the de-
liberate process, that authority does not completely
resolve the dilemma.
2. Cleanup is taking place within the framework
of legal and regulatory procedures. The XRC.
aided by public comments, is preparing a compre-
hensive environmental impact statement, which
will set forth a range of alternatives for accom-
plishing cleanup and will consider their environ-
mental effects. 262 In addition, separate environ-
mental assessments have been prepared by the
XRC, and circulated for public comment, to assess
whether certain specific cleanup steps thus far
proposed would have significant adverse environ-
mental consequences. 263
The XRC is also reviewing proposed modifica-
tions to the Unit 2 license. Its procedures provide
for formal hearings involving the licensee and
parties who may be affected by the modifica-
tions. 264 At issue will be : (a) whether the proposed
modifications are necessary and sufficient for the
maintenance of the facility and for the protection
of public health and safety: and (b) whether they
would significantly affect the quality of the en-
vironment. 265
The various procedures are intended to afford
orderlv decisionmaking that provides an oppor-
tunity for some form of input from interested
parties, including members of the public, particu-
'Pp. 194-195.
' Pp. 197-198.
'P. 201. ""P. 201.
1 P. 208.
J "P. 196. ""P. 196. M 'P. 196.
151 Pp. 197-199. '"P. 198. "P. 198.
"P. 201. '"P. 200.
148 Pp. 196-200. ""Pp. 197-198. and fn. 114. p. 198.
** P. 198. " P. 199. ** P. 198. " Pp. 199-200.
Pp. 201. 204-205.
'Pp. 201-204, 205-207. "*P. 208.
21
larly on environmental matters. 266 They alsjo a^e
intended to help resolve differences among the
parties involved in or affected by .cleanup. These
decisionmaking processes are by nature delibera-
tive and extend the time required for reaching
final decisions.
3. Legal and regulatory procedures do not
necessarily prevent the NRC from taking immedi-
ate action that otherwise would await an environ-
mental review. Both the NRC and the Council on
Environmental Quality (CEQ) have agreed that
if there is an "emergency circumstance," the NRC
is authorized to take prompt, specific action even
before completion of the comprehensive environ-
mental impact statement.
In non-emergency situations, the NRC. but not
necessarily CEQ, maintains that the NRC may
take prompt, specific action before completion of
the impact statement when necessary to protect
public health and safety and so long as an assess-
ment has been made, with public comment, that the
particular action will not have a significant ad-
verse environmental impact. 267
Any differences between the two agencies should
be resolved promptly on what may be done in non-
emergencies.
4. In October 1979 the NRC authorized prompt
use of EPICOR-II to clean radioactive water in
the auxiliary building based on the NRC's finding
that such action was necessary to protect public
health and safety and would have no significant
adverse environmental impact. The CEQ, con-
curred but only after concluding that there was an
"emergency circumstance." 268
In June 1980, the NRC authorized prompt vent-
ing of the containment, finding that it was in the
best interest of public health and safety, would not
have a significant adverse environmental impact,
and would not limit the choice of reasonable alter-
natives for future cleanup steps. In those circum-
stances, CEQ concluded the action would not vio-
late applicable Federal Regulations.
5. The investigation found problems relating to
the NRC's planning and management of cleanup
that cannot be attributed to the deliberate pace
of legal and regulatory procedures.
In early November 1979, some seven months
after the accident, the NRC had not formulated
new and specific regulatory guidelines to govern
radiological releases during cleanup activities ; nor
had it permitted the licensee to follow the existing
' regulations, which applied to normal plant oper-
ations. 269 Certain cleanup steps thus were reviewed
on a case-by-case basis without any clear indica-
tion of what radiological releases, if any, would
be acceptable. 270
In addition, even though the issue had been
raised within two months of the accident, the NRC
took until November 21 to decide whether to pre-
pare a programmatic environmental impact state-
ment, losing many months of valuable time.
Finally, at Subcommittee hearings in early No-
vember, NRC officials offered no specific schedules
for how long cleanup would or should take to
complete. 271
More than three months later, problems in these
areas persisted. In late February 1980, the NRC
Special Task Force concluded that interim cri-
teria were needed to permit radiation releases
associated with plant maintenance and data-
gathering activities because, lacking any criteria
whatsover, NRC staff had tended to submit for
prior approval by the Commissioners every
cleanup proposal that did not meet a "zero re-
lease" requirement. 272 The task force also found
that although completion of the environmental
impact statement was an important milestone in
the cleanup, the staff still was not clear as to how
the Commissioners intended to use the state-
ment. 273 Finally, the task force found that cleanup
schedules were needed, noting that over the
months both the NRC and the licensee had begun
giving less priority to developing and implement-
ing cleanup plans. 274
Based on the task force's recommendations, in-
terim criteria for releases finally were prepared. As
of May 1980, however, the NRC still had to deter-
mine how and by whom major cleanup decisions
would be made after the environmental impact
statement is completed. The Commission also still
had to determine the role this statement would play
in making these important decisions. 275 The Com-
mission had to decide, for example, whether the
agency will insist on those cleanup proposals that
are believed to have the smallest adverse environ-
mental impact, whether it will set explicit regu-
latory guidelines to govern radiological release
during cleanup and who will have authority to
give final approval to proposed steps as cleanup
proceeds. These are decisions that must be made.
""Pp. 19S, 202. 205-208.
169-170. Pp. 169-171.
*"Pp. 203. 206. "'Pp. 203. 207. M9 Pp. 169-170.
ln Pp. 171, 204. m P. 171. ** Pp. 171, 204.
'Pp. 169-170, 206.
'Pp.
22
Chapter 3
How The Plant Works
23
%
Reactor Coolant
Pumps (4)
Once-Through
\\ Steam Generators (2)
Pressurizer
Reactor Vessel
Adaptation from, Babcock & Wilcox diagram
The nuclear steam supply system at Unit 2, Three Mile Island
24
Chapter 3
How The Plant Works
NUCLEAR VS. NON-NUCLEAR PLANTS
To understand how a nuclear powerplant
works, two important points should be kept in
mind.
First, a nuclear plant is very similar to a con-
ventional coal- or oil-fired powerplant. In both,
water is heated to produce steam. The steam turns
a turbine that drives a generator to produce elec-
tricity. Each type of plant has a large, elaborate
plumbing system to heat the water, carry steam to
the turbine, condense the steam back into water,
and then return the water to the source of heat
similar to the way plumbing in a house carries
heated water from a furnace to the radiators and
back to the furnace for reheating and recirculation
to the radiators.
Second, a nuclear powerplant is very different
from a non-nuclear plant in certain essential fea-
tures. The plumbing in a nuclear plant serves as a
safety system that is not needed in a fossil-fueled
plant. Nuclear fuel is a ceramic made from ura-
nium, a metal, that produces much more intense
heat than does fossil fuel. It must be kept covered
by rapidly moving water, or coolant, that removes
the heat and keeps the temperature of the fuel
below its melting point. Molten nuclear fuel has
the potential to penetrate a plant's structure and
foundation and to cause hazardous offsite releases
of radioactivity. Even after a nuclear plant is shut
down, the fuel produces considerable heat enough
to melt the fuel and must be cooled for a sub-
stantial period by circulating water.
The plumbing in a nuclear plant, therefore, pro-
vides a series of redundant safety systems to ensure
that the fuel is constantly covered with water. Fos-
sil fuel does not continue to produce large amounts
of heat after the fire is stopped and, therefore, does
not require this kind of cooling.
THREE MILE ISLAND, UNIT 2
The Three Mile Island Nuclear Power Station
has two nuclear realtors, Unit 1 and Unit 2, each
capable of delivering about 880 million watts
(880 megawatts) of electricity, enough to serve a
city of nearly 2 million people.
Each reactor is a pressurized-water type, mean-
ing that pressure within the reactor and the pipes
leading to and from it is kept high at about 2.200
pounds per square inch. By maintaining this high
pressure, water running through the reactor is
prevented from boiling at the usual boiling point
of 212 Fahrenheit. This permits the water to be
heated to much higher temperatures and still be
kept in a liquid state without steam bubbles. This,
in turn, permits the plant to produce steam far
more efficiently than if the water were to boil at
normal, atmospheric pressure. (Another type of
reactor, the boiling water reactor, is less pressur-
ized 1,000 psi and produces steam directly from
the boiling water in the reactor.)
THE BUILDINGS
The reactor is the heart of any nuclear power-
plant. At Three Mile Island Unit 2, the reactor is
housed inside a massive, domed structure known as
a containment building* also referred to as the
reactor building. This structure, rising 193 feet
above the Susquehanna, has steel-lined, reinforced
concrete walls almost two feet thick. The contain-
ment provides the final line of defense against
escape of high levels of radioactivity from inside
the reactor. The containment building also holds
some of the major elements of the plant's nuclear
25
5U-058 0-80-3
VENT STACK
CONTAINMENT BUILDING
DISCHARGE LINE
' PILOT-OPERATED
CODE SAFETY VALVE^ T RELIEF VALVE (PORV)
i BLOCK VALVE
STEAM GENERATOR
B"
AUXILIARY
BUILDING
REACTOR VESSEL
REACTOR
COOLANT PUMP
BORATED
WATER
STORAGE
TANK
steam supply system a massive array of pipes,
pumps, tanks and valves for circulating coolant
through the reactor, and a pair of steam gener-
ators, each one a 73-foot tall cigar-shaped struc-
ture in which steam to drive the turbine is pro-
duced. In addition, the containment building
houses portions of the Emergency Core Cooling
System (ECCS), which ensures an adequate sup-
ply of water to the nuclear fuel in the event of an
accident.
The containment is but one of several buildings
and structures that comprise Unit 2. Only one
building is shared by the two TMI units the fuel
handling building, where relatively non-radio-
active fresh fuel is stored without shielding before
being loaded into the two reactors. It is also where,
after being removed from the reactors, the highly
radioactive spent fuel is stored in steel-lined
"swimming pools'* beneath 40 feet of water.
Unit 2 has an auxiliary building where large
pipes, pumps, tanks and filters help to maintain
the level and purity of the water flowing through
the nuclear steam supply system in the adjacent
containment building. This building also contains
portions of the Emergency Core Cooling System.
There also is a turbine building where the main
steam line from the steam generators in the con-
tainment connects with the turbine to drive the
electricity-producing generator. Here the steam
is also cooled and condensed into water and the
water purified of minerals before being returned
to the steam generators in the containment
building.
Outside the Unit 2 turbine building stands a
26
MAIN CONDENSATE
FEEDWATER BOOSTER
PUMP PUMP
CONDENSATE
STORAGE
TANK
Adapted from: The Report of the President's Commission on the Accident at Three Mile Islar
Schematic of principal system* and components, Unit 2
pair of 350-foot tall cooling towers the now-
familiar hyperboloid structures from which
plumes of vapor rise, a product of the process of
condensing the steam that has passed through the
turbines.
Finally, there is the control building in which
the control room, the nerve center of each plant, is
located. It is from here that operators monitor
and control the operations of vital plant equip-
ment to ensure that heat is being removed effec-
tively from the reactor.
THE REACTOR
The reactor is a nuclear furnace in which ura-
nium fuel gives off intense heat, leading to fuel
temperatures of as much as 3.250 F under normal
operating conditions. The heat is produced by
nuclear fission, the same splitting of uranium
atoms in a chain reaction that takes place in nu-
clear weapons and that has been known to exist in
nature. But it happens at a slower, controlled
rate in nuclear powerplante. so that it is impos-
sible for them to experience nuclear explosions.
This is partly because the reactor's fuel is in a
dilute form known as low-enriched uranium. Even
in the worst conceivable accident, there cannot be
an atomic explosion.
The reactor in the TMI-2 plant has several com-
ponent parts. It is encased in a 36-foot high tank
with steel walls nearly nine inches thick. This tank,
known as the reactor vessel, is in turn encased by
9 l /2 feet of steel and concrete in the form of two
separate shields. The top of the vessel, the reactor
27
head, is removable to allow for refueling. The re-
actor vessel and its shielding provide the inter-
mediate line of defense against radioactive releases
from the fuel inside the reactor core.
THE CORE
The core at TMI-2 holds almost 100 tons of
uranium within 177 fuel assemblies. Each fuel as-
sembly holds 208 fuels rods thin, 12-foot-long
metal tubes containing the uranium fuel. The fuel
inside the rods is a compressed powder known as
uranium oxide that is molded into ceramic fuel
pellets. Each pellet is about an inch long and less
than half an inch wide ; they are stacked inside the
fuel rods, which, in turn, are grouped into the fuel
assemblies. In all, there are 36,816 fuels rods in
the reactor core.
The fuel rods, which are made of an alloy of
the metal zirconium, known as Zircaloy, serve
three purposes. First, they provide the initial line
of defense against the potential release of hazard-
ous radioactive materials, known as fission prod-
ucts, that form in the uranium fuel when the
Reactor Head
Control Rods and
Control Guide Tubes
Coldleg
Core
Fuel Assembly
(Includes Fuel Rod)
Fuel Pellets
in Fuel Rod
Guide Tube
Fuel Rod
Zircaloy
Cladding
Pathways for
Incore Instrumentation
Fuel Assembly
Reactor Vessel and Core
(vessel filled with coolant)
Adapted from: Babcock & Wilcox
28
Adapted from: Babcock & Wilcox
reactor is operating. The products are contained
within the Zircaloy walls, or cladding.
Second, the Zircaloy cladding permits the almost
unobstructed passage of atomic particles called
neutrons, which, when jettisoned in the splitting of
uranium atoms, strike other atoms, causing them
to split apart the so-called chain reaction.
Finally, the fuel rods promote the transfer of
heat from the fuel to the coolant water being
pumped through the core.
The nuclear fission process inside the reactor is
controlled by the insertion of control rods into the
fuel assemblies and by the addition of boron into
the coolant. The control rods are long tubes shaped
like the fuel rods. They contain materials that
absorb neutrons. These materials, known as
"poisons" include indium and cadmium.
During normal, full-power operations, the con-
trol rods are withdrawn from the core. The rate
of the chain reaction is then controlled by boron
in the coolant, the amount of which can be ad-
justed. Boron, too. absorbs neutrons.
During an accident, or any sequence of events
that seriously interferes with the normal removal
of heat from the core, the control rods will auto-
matically drop all the way into the core, thereby
"tripping" the reactor and instantaneously termi-
nating the chain reaction. This, in turn, stops
most of the heat generation by the core, although
considerable heat, called decay heat, remains.
THE PRIMARY SYSTEM
Normally, a nuclear powerplant operates with
marvelous precision on a massive scale. Water
flows through the core at a rate of 92,400 gallons
a minute, pushed by four reactor coolant pumps
each five stories high and 9,000 horsepower. Under
normal operating conditions the water is heated to
nearly 600F and is subjected to some 150 atmos-
pheres of pressure (2.200 pounds per square inch,
equivalent to pressures nearly a mile deep in the
ocean). The water leaves the reactor through two
pipes, each three feet in diameter, known as the
''Kotlegs*' 1 One hotleg leads to steam generator A
the so-called "A loop" the other to steam generator
B. the "B loop" This system for circulating water
through the core is known as the primary system.
THE PRESSURIZER
Pressure in the primary system is maintained
and fine-tuned by a 42-foot-high tank known as the
l>f< xxur'izer. In some ways, the pressurizer is like an
expansion tank in a home hot water heating sys-
tem: it provides a place for water in a closed
plumbing system to collect when it expands after
being heated. An expansion tank, however, is a
passive device that simply collects excess water,
whereas the pressurizer actively controls pressure
in the primary system.
The pressurizer at TMI-2 normally holds 800
cubic feet of water, on top of which is a cushion,
Rot-operated
Relief Valve (PORV)
Block Valve
Code Safety Valves
Spray
Steam Bubble
-Normal Water Level
Heater
Water Inlet
The Pressurizer
Adapted from: Nuclear Safety Analysis Center
or "&M&&&," of 700 cubic feet of steam. Pressure is
controlled in the rest of the system by expanding
and contracting the steam bubble, which pushes
against the primary system water at the bottom
of the tank. The bubble is expanded by means of
heaters in the tank that produce more steam, in-
creasing pressure ; or it is diminished by means of
sprays that condense some of the steam into water,
thereby lowering pressure.
If the bubble is lost while the reactor is operat-
ing, it is extremely difficult to control pressure
in the primary system. Sudden increases in pres-
sure could damage the primary system or break
primary piping, since there would be no bubble
serving as a buffer. The bubble can be lost if too
much water gets into the pressurizer. Operators
29
are trained to avoid having the pressurizer "go
solid," as a pressurizer full of water is called.
If pressure in the reactor rises so rapidly that
the pressurizer sprays cannot counteract it, a re-
lief valve at the top of the pressurizer, known as
the pilot-operated relief valve, or PORV, opens
automatically. Steam is released through a dis-
charge line that leads to a reactor coolant drain
tank on the floor of the containment. The PORV
is designed to close automatically as pressure in
the primary system returns to normal.
There is a back-up safety system that comes
into play if additional pressure must be relieved,
or if the PORV fails to open or has been "isolated"
by a block valve because it is leaking. This system
involves what are known as code safety valves.
They open automatically on high pressure and
close automnticallv as normal pressure is restored.
Unlike the PORV, the code safety valves cannot
be isolated, thpt is. blocked on command from the
control room. They are intended to serve as the
final line of defense against excessive pressure in
the primary system.
THE STEAM GENERATORS
Under normal operating conditions at TMI-2,
the heated, pressurized water in the A and B
primary loops passes through the hotlegs. which
have "candy c<z7)e"-shaped curves at their high
point, and then enters the corresponding A and B
steam generators.
This water transfers some of its heat to cooler
water that enters the steam generators from a
separate closed system the feedwater system on
the secondary side of the plant. The water on
the primary side, which is radioactive, passes
through the steam generators in a series of long,
narrow tubes, around which the non-radioactive
secondary system water flows. The radioactive
primary system water leaves the bottom of the
steam generators via pipes known as "coldlegs"'
and is pumped back into the reactor for reheating
and recirculation to the steam generators.
THE SECONDARY SYSTEM
The non-radioactive water in the steam genera-
tor boils and turns to steam after being heated by
radioactive coolant water from the reactor. The
non-radioactive steam leaves the steam generators
through the main steam lines and travels out of
the containment and into the turbine building,
where it enters the turbine. The turbine drives the
generator, which produces electricity. Steam from
the turbine enters a condenser, where it is con-
densed into water. A condensate pump then pushes
this water through a condensate polisher unit that
purifies it in a manner similar to the way a home
water-softener works.
A condensate booster pump then moves the
purified water to the main feedwater pump that, in
turn, pushes the water back into the secondary
side of the steam generator, where it is boiled into
steam again for recycling to the turbine.
CONDENSER COOLING SYSTEM
The condenser is cooled by water from yet
another closed system. This water, which absorbs
heat from the steam in the condenser, is pumped
from the condenser to the cooling towers. There it
cascades down a series of steps, giving up heat
which appears as vapor clouds rising into the
sky. This vapor is not radioactiA^e.
Water from the cooling towers is pumped back
to the condenser, where the cycle is repeated.
THE SAFETY SYSTEMS
TMI 2, like other pressurized water reactors,
has elaborate and redundant safety systems on
both the secondary and primary sides of plant to
assure adequate cooling of the core.
A LOSS OF FEEDWATER
On the secondary side, a loss of feedwater to the
steam generators is a potentially serious problem
because the steam generators soon would run dry,
thus eliminating the principal means of removing
heat from the primary system. This, in turn, would
cause temperature and pressure in the core and
elsewhere in the primary system to rise rapidly.
In the event of a loss of feedwater caused by a
broken pipe, failed pump or other malfunction on
the secondary side of the plant, there is a set of
emergency feedwater pumps that can provide an
alternative supply of water from a condensate
storage tank.
However, the emergency feedwater pumps can
be overriden by shutting a set of valves known as
the "No. 12 valves" that block the flow from these
pumps to the steam generators. Inexplicably, these
valves were closed at TMI at the start of the ac-
cident on March 28, 1979.
In the event the flow of feedwater to the
steam generators cannot be maintained, then the
supply of steam to the turbine cannot be main-
tained either. The turbine will automatically react
to this problem when the feedwater pumps trip.
The turbine will then trip that is, shut down
to avoid damage.
On the primary side, if conditions depart suffi-
ciently from the norm, the reactor will automati-
cally trip by dropping its control rods all the way
into the core, thus terminating the chain reaction.
This is also known as a "scram"
30
A LOSS OF COOLANT
The sudden increase in temperature and pres-
sure prior to the scram may cause the PORV to
open briefly, but, as noted, it is designed to close
as pressure drops back to normal. If the PORV
should remain open without being detected by con-
trol room personnel, as it did at the start of the
TMT accident, then the primary system will lose
coolant through the pressurizer a potentially
serious "small '-break" in the system, resulting in
a I oss-of -coolant accident (LOG A). If sufficient
coolant were lost without being replenished, the
core could become uncovered, and severe damage
could result, including melting of the fuel.
Again, there is a safety system the plant's
Emergency Core Cooling System. It consists of
several back-up safety subsystems designed to
compensate for small-break LOCAs, such as leak-
age through the pressurizer, or even for a large-
break loss-of -cool ant accident, such as a rupture
of the three-foot-wide coldleg or hotleg pipes.
SMALL-BREAK LOCAS
In the event of a small break in the primary sys-
tem, additional coolant is provided by the auto-
matic start-up of the high pressure injection
( HPI) *>i*tfm. It uses the make-up pumps, located
in the auxiliary building, that are normally used
to replenish the primary system through the high
pressure injection of berated water into the cold-
legs. The source of this additional water is the
Borated Water Storage Tank. This emergency sys-
tem operates when pressure in the primary system
is high, the case with small-break loss-of-coolant
accidents, in which little pressure is lost because,
as the name implies, the break is small.
LARGE-BREAK LOCAS
In the event of a large break in a coolant pipe,
pressure would drop so low that the high pressure
injection system would be supplemented by other
parts of the Emergency Core Cooling System.
Core -flood tanks directly above the reactor would
dump thousands of gallons of coolant directly into
the reactor vessel. They drop their water onto the
core as soon as reactor pressure drops below 600
psi
As pressure drops further, a low pressure in-
jection (LPI) system (not shown in the figure),
also drawing from the Borated Water Storage
Tank, provides coolant at a much higher rate.
If the supply of water in the tank is depleted,
water may be drawn from the containment sump,
where water flowing out the break will collect.
DECAY HEAT REMOVAL SYSTEM
After a reactor scram, residual or decay heat
must continue to be removed from the core. This is
the heat generated by the radioactive decay of
fission products in the nuclear fuel even after the
chain reaction has been halted. This decay heat is
substantial substantial enough to melt the fuel if
the core is not kept covered with coolant water.
But it diminishes rapidly at a steady rate over a
period of several hours to a relatively low level, but
still substantial amount, of heat.
With the Emergency Core Cooling System keep-
ing the core covered, plant operators work to bring
primary system temperature and pressure down bj-
removing heat through the steam generators.
When temperature is reduced to about 300F and
pressure to about 400 psi, low pressure injection
pumps would be used to circulate coolant. In this
case, the coolant goes not to the steam generators,
but to a separate heat exchanger located outside the
containment. The LPI pumps (when used in this
manner), the heat exchangers, and the piping are
known as the decay heat removal system. This sys-
tem permits temperature to be lowered below the
boiling point of 212F to about 120F and
depressurization to atmospheric pressure. At that
point the plant is in a stable state known as cold
sh utdown.
31
Chapter 4
How The Accident Happened:
A Mechanical Summary
33
FROM NORMAL CONDITIONS
I I Primary Water
Secondary Water
I I Steam
FA \ Steam/Hydrogen
Steam generator
Return to
reactor vessel
Loop A Loop B
1. Coolant throughout the primary system; core completely covered.
TO SATURATED STEAM .
Pilot-operated
relief valve (PORV)
^x~ Block valve
r*.-^l Primary Water
Secondary Water
I I Steam
I." J Steam/Hydrogen
Loop A Loop B
2. Coo/ant lost through the stuck-open PORV; decreased pressure caused coolant to boil; reactor coolant
pumps had to be turned off; saturated steam rose out of coolant; core barely covered with coolant.
(Continued on page 36)
34
Chapter 4
How The Accident Happened:
A Mechanical Summary
THE FIRST SECONDS
The nuclear accident at Unit 2 of Three Mile
Island liegan 36 seconds after 4 a.m. on March 28,
1979. when all the outlet valves on the condensate
water polishing system closed, tripping the feed-
water pumps. This, in turn, stopped the flow of
water to the steam generators on the secondary
side of the plant. At that point, the turbine tripped.
The emergency feedwater pumps activated auto-
matically to maintain flow to the steam generators,
but. inexplicably, the valves were closed between
the pumps and the steam generators, blocking the
flow. As a result, no water on the secondary side
could reach the steam generators. 1
All this occurred in the first seconds. Heat in the
reactor vessel and the rest of the primary system
began to increase, causing a rapid rise in pressure
in the primary system. This, in turn, caused the
pilot-operated relief valve (PORV) on the pres-
surizer to lift. Pressure continued to rise. Very soon
the reactor tripped, and the control rods fell into
position between the fuel rods. Pressure in the
primary system began to fall as less heat was
generated in the reactor. The accident was still
only seconds old.
A LOSS OF COOLANT ACCIDENT
The PORV failed to close, as designed, when
the pressure dropped. Steam and water continued
to flow, undetected, out of the pressurizer.
Pressure in the primary system continued to fall
as the volume of coolant contracted from the loss
of heat and as coolant escaped through the stuck-
open PORV.
A Io5?-of-coolant accident (LOCA) was under-
way. It went undetected because control room per-
sonnel did not realize the PORV was stuck open.
Two minutes into the accident, the high pres-
sure injection system (HPI), an emergency sys-
tem designed to compensate for a loss of coolant,
automatically started pumping water into the re-
actor vessel at 1,000 gallons per minute. Mean-
while, the pressurizer was filling with water. In
response, operators severely throttled this flow to
avoid overfilling the pressurizer.
The limited amount of water flowing into the
primary system was inadequate to replace the
amount being lost through the PORV a poten-
tially dangerous loss of coolant if not corrected.
STEAM IN THE SYSTEM
Within minutes, as a result of the loss of pres-
sure in the primary system, the coolant began to
boil, causing saturated steam to form in the cool-
ant. At about one hour into the accident, the reac-
tor coolant pumps that circulate the water through
the primary system began vibrating because they
were beginning to pump the steam-water mixture
produced by the boiling. At about 114 hours, two
were turned off to prevent damage; the last two
were turned off at 1% hours into the accident.
CONDITIONS WERE NOT UNDERSTOOD
For the first 1% hours, control room personnel
struggled to understand what was happening in
the plant. Hundreds of alarms went off, signaling
such things as unusual conditions in the reactor
coolant drain tank, high temperature and pressure
in the containment building, and low pressure in
the primary system. The conditions that developed
were beyond those that control room personnel had
experienced in their training or in their operation
of the plant. The symptoms described in the emer-
gency procedures did not exactly fit the situation
and proved of little help.
1 Eight minutes later an operator opened the valves after discovering they were shut.
THEN CORE UNCOVERING AND SUPERHEATED STEAM
- Pilot-operated
relief valve (PORV)
- Block valve
I I Primary Water
BB Secondary Water
I I Steam
I " J Steam/Hydrogen
Loop A
Loop B
3. Core becoming uncovered, exposed fuel heating up; steam in system became superheated.
AND HYDROGEN
Pilot-operated
relief valve (PORV)
Block valve
l"~~l Primary Water
BB Secondary Water
I I Steam
rr 7 ! Steam/Hydrogen
Loop A
Loop B
4. Core uncovering continuing; temperatures in core hot enough that hydrogen was generated as a result
of a chemical reaction betiveen superheated steam and the Zircdloy fuel cladding ; hydrogen and
superheated steam collecting in hotlegs.
36
CORE UNCOVERING
Around 5 :45 a.m.. very soon after the shutdown
of the last two reactor coolant pumps, the core
became uncovered. The uncovering of the core oc-
curred because, with the pumps off, steam gener-
ated by the boiling in the core rose to the higher
portions of the reactor vessel and the rest of the
primary system, while water continued to escape
from the kuck-open PORV and while the HPI
remained throttled. Water level fell below the top
of the core.
Over the next half hour, the water level fell
further until the top two-thirds of the core was
exposed. Fuel rods crumbled. Hydrogen was pro-
duced as steam reacted chemically with the Zirc-
aloy fuel cladding. Fission products escaped from
the failed fuel into the coolant of the primary
system.
Plant operators and managers still did not
realize the core was uncovered. They were unaware
of the stuck-open PORV. and they had no direct
means of measuring the level of water in the core.
Finally, at about 6:20 a.m.. a shift supervisor
who had arrived in the control room about a half
hour earlier realized the PORV was stuck open.
He ordered that a backup valve, the block valve, be
closed. It was. and the loss of coolant was 'stopped.
A SITE, THEN GENERAL EMERGENCY
At 6:45 a.m.. a site emergency was declared,
based on radiation levels inside the plant. At 7 :24
a.m.. a general emergency was declared, based on
the potential for offsite radioactive releases. The
utility notified State and Federal officials when it
declared the site and general emergencies.
Shortly before 7 :30, flow from the high pressure
injection system was increased. The core was even-
tually covered again. But steam and hydrogen gas
had become trapped in the hotlegs of the primary
system, blocking circulation of water through the
system.
By this time, three and a half hours into the
accident, most of the damage to the core had been
done, and radiation levels in the plant were high.
STRATEGIES TO REACH STABILITY
For the rest of the day, control room personnel
struggled to regain stability in the plant. The prin-
cipal problem was to ensure a reliable flow of water
through the core. In the morning hours, they first
tried to repressurize the system in order to collapse
what they believed to be saturated steam bubbles in
the system. The blockage was actually caused by a
mixture of superheated steam and hydrogen,
neither of which could have been condensed into
the coolant.
With the failure of repressurization, concern
arose over whether the core was covered and
whether the limited supply of HPI water available
would become exhausted. These uncertainties led to
the next strategy depressurization of the primary
system. Utility personnel reasoned that lower
pressure would activate the core flood tanks, which
would dump more water onto the core, assuring
that it would be covered.
WAS THE CORE COVERED?
At about 11 :30 a.m. the block valve was opened,
allowing steam and gas once again to escape from
the pressurizer. Pressure dropped. The core flood
tanks eventually dumped water onto the core, but
only a limited amount* Some control room person-
nel interpreted this to mean the core was covered ;
others concluded that the core had never been
uncovered.
Confident the core was covered, at 1 :10 p.m.
plant operators and managers halted depressuri-
zation.
THE HYDROGEN BURN
About 40 minutes later, two members of the
emergency command team decided to depressurize
again in the hope of reaching a low enough level of
pressure to permit use of the low pressure decay
heat removal system. As the block valve was
opened, there was an extremely sharp increase in
pressure and temperature in the containment, ac-
companied by activation of the containment
sprays. This happened when hydrogen in the con-
tainment ignited. The hydrogen which had been
generated by a chemical reaction between the
cladding of the fuel and the steam, burned only a
few seconds.
Depressurization again was unsuccessful. For
reasons still not definitey understood, pressure in
37
the primary system could not be lowered to the
point at which the decay heat removal system
could be initiated. During this time, the core may
have been uncovered again.
STABILITY ACHIEVED
Finally, about 5 :30 p.m., utility executives off-
site ordered the emergency command team to re-
pressurize the system again. The objective was to
collapse enough steam in the primary system to
permit the restart of a reactor coolant pump. This
time the strategy worked. At 7 :50 p.m., relatively
stable conditions were achieved as the pump
started circulating water through most of the core
and the rest of the primary system.
All the damage to the core occurred on the first
day. More crises followed, with discovery of the
damage to the core on the third day and the ensu-
ing uncertainty caused by the now-famous hydro-
gen bubble.
Finally, several days later, natural circulation
in the primary system was finally achieved.
38
CHRONOLOGY OF EVENTS, MARCH 28, 1979
Following is a brief chronology of the major events of the accident during the first day :
Brief chronology of events? March 28, 1979 Brief chronology of events,* March 28, 1979
Elapsed time
since the acci-
Clock time dent began Event
4:00 a.m-.- 00:00:00
Do..
Do--
Do--
Do..
4:02 a.m.
4:05 a.m.
4:06 a.m.
00:00:01
00:00:03
00:00:08
00:00:13
00:02:02
00:04:38
00:05:30
4:08 a.m.- 00:07:29
Do.. . 00:08:18
4:11 a.m 00:11:00
4:15 a.m.. . 00:15:00
Lost of feedwater.
Initiated the accident ; emergency
feedwater system starts but
fails to supply the steam gen-
erators because of closed
valves.
Turbine shuts off.
Automatic upon loss of feed-
water.
PORV opens.
Relieved high primary system
pressure; provides path for
loss of coolant.
Control rods drop.
Stops fission process, but decay
heat still must be removed.
PORV fails to reclose.
Mechanical failure of the valve
resulting in continued loss of
primary coolant; plant per-
sonnel do not realize valve is
still open.
High pressure injection initiated.
Automatic upon low primary sys-
tem pressure.
High pressure injection throttled 1 .
Throttled back to maintain con-
stant pressurizer level.
Saturation conditions in primary
system.
First steam bubbles form in the
primary system.
Pumps start sending water to aux-
iliary building.
Automatic with high water level
in the containment sump;
water only slightly contami-
nated.
Emergency feedwater valves opened.
Plant personnel notice closed
valves; opened to initiate flow
to steam generators.
High containment sump level
alarm.
Abnormal amounts of water pres-
ent in containment.
Reactor coolant drain tank rup-
tures.
Flow from PORV ruptures tank;
water spills onto containment
floor.
Elapsed time
since the acd-
Clock time dent began Event
4:20 a-m 00:20:00
4:38a.m 00:38:00
5:14 a.m__
5:41 a.m..
01:14:00
01:41:00
5:45a.m . 01:45:00
6:22a.m.
6:56a.m-
7:20 a.m-
7:24 a-m-
02:22:00
02:56:00
03:20:00
03:24:00
7:56 a.m . 03:56:00
8:26 a.m 04:26:00
9:15 a.m . 05:15:00
11:38 a.m 07:38:00
12:41 p.m 08:41:00
Abnormal neutron flux behavior.
Instruments measuring neutron
flux begins reading abnormally
high.
Pumps that send water to auxiliary
building shut off.
Water retained in containment
sump after about 8,000 gallons
of slightly radioactive water
pumped to the auxiliary
building.
| Reactor coottant pumps turned off.
Essentially, flow through the core
stops.
Initial core uncovering begins.
Water level drops and heat re-
moval is diminished; fuel
damage results.
Block valve for PORV closed.
Loss of coolant halted.
Site emergency declared.
Because of high radiation; NRC
and State officials notified.
High pressure injection increased.
Operators initiate increased high
pressure injection flow.
General emergency declared.
Because of high radiation; NRC
and State officials notified; off-
site radiation monitoring teams
dispatched.
Containment automatically isolated.
High containment pressure initi-
ates automatic isolation to
prevent radiation release.
Sustained high pressure injection.
From this time on, high pressure
injection is continuously main-
tained, at varying flow rates,
after having been turned off
altogether for about 5 minutes.
Initial repressurization.
Attempt to collapse vapor bub-
bles in the system and establish
natural circulation.
Depressurization.
Operators open PORV block
valve to reduce pressure and
inject water from core flood
tanks to assure themselves that
core is covered.
Core flood tanks initiated.
Little water injected; plant per-
sonnel believe that this indi-
cates core is covered.
1 Footnote at the end of table.
1 Footnote at the end of table.
39
Brief chronology of events, 1 March 28, 1979
Brief chronology of events? March 28, 1979
Elapsed time
since the acci-
Clock time dent began Event
1:10 p.m. 09:10:00
1:50 p.m.-. 09:50:00
Depressurization halted.
Convinced core covered, plant
personnel close the PORV
block valve, halting further
depressurization.
Second depressurization and con-
tainment pressure "spike."
Operators open the PORV block
valve to depressurize to allow
use of the decay heat removal
system. Simultaneously, a con-
tainment pressure spike occurs
because of the combustion of
hydrogen in the containment.
Elapsed time
since the acci-
Clock time dent began Event
3:08 p.m... 11:08:00
5:20 p.m... 13:20:00
7:50 p.m... 15:50:00
Depressurization ends.
Operators close PORV valve,
ending attempts to depressur-
ize further. They failed to
reach pressure for decay heat
removal system.
Repressurization.
Attempt to collapse vapor bub-
bles and establish forced cir-
culation using reactor coolant
pump.
Reactor coolant pump started.
Forced circulation through core
and relatively stable plant
conditions established.
1 All times are approximate.
40
Chapter 5
Radiation Effects And Monitoring
41
5-4-358 O-80-i*
Helicopter monitoring radiation releases during March 28, 1979 accident
42
Chapter 5
Radiation Effects And Monitoring
The foremost concern in the event of an acci^
dent at a nuclear power plant is the amount of
radioactive material that may escape and its ad-
verse health effects on plant workers and the sur-
rounding population.
MEASURING RADIATION
TYPES OF RADIATION
All life is constantly exposed to natural and
manmade radiation that is transmitted in such
common forms as risible and invisible (infrared}
light, radion-ares and microwaves. X-rays and
cosmic rays. There are two types of radiation
the "non-ionizing" type, as produced by micro-
wave ovens, and the "ionizing" type, as produced
by radioactive materials such as those used and
produced by nuclear power plants.
Radiation, in its passage through matter, can
activate atoms to generate heat but still leave the
basic structure of the atoms unaltered in the
process. This is characteristic of non-ionizing
radiation. Ionizing radiation, on the other hand,
can alter the atomic structure by knocking a nega-
tively charged electron from an atom, leaving
behind a positively charged atom known as an
"ion" hence the name "ionizing radiation." These
ions can be produced in molecules found in the cells
of living tissue. Since normally functioning cells
depend on a delicate electro-chemical balance, the
presence of ions within cells can cause harm to the
body. As described below, the extent of cellular
damage and bodily harm depends on the tvpe of
ionizing radiation and the amount absorbed by the
bodv.
Radioactive materials such as uranium produce
two basic types of ionizing radiation : one in the
form of atomic particles (alpha and beta parti-
cles} . known as particulate radiation ; the other in
the form of electromagnetic energy (gamma rays
and X-rays} . known as electromagnetic radiation.
CHARACTERISTICS OF RADIATION
The comparatively heavier alpha particles
travel only a few inches in the air and cannot pene-
trate the skin. However, they are hazardous if the
radioactive material producing alpha particles is
breathed or eaten. Then these particles can cause
intense damage to nearby cells.
The smaller, lighter beta particles are more
penetrating, travel greater distances and can pene-
trate the upper layers of the skin.
Gamma and X-rays take the form of energy
moving at the speed of light. Gamma rays are
more energetic than X-rays and can penetrate
deeper; they can be used to take "photographs"
through such relatively impenetrable material as
steel. Gamma rays and X-rays, unlike alpha and
beta particles, can penetrate the body from outside
and damage tissue deep within the body.
PRODUCTION OF RADIATION
Radioactive materials emit one or more types
of radioactive particles or energy over various
periods of time, eventually losing their radioac-
tivity. The overall rate of decay of these materials
into non-radioactive forms is measured in terms of
the half-life of the material that is, the amount
of time it takes one-half of the atoms in the mate-
rial to decay and become non-radioactive.
At a nuclear power plant, a large number of
radioactive materials are produced by the fission-
ing of the uranium in the nuclear fuel. The half-
lives of these "-fission products" range from sec-
onds to hundreds of millions of years. These prod-
ucts in turn produce alpha and beta particles,
gamma rays and X-rays.
UNITS OF MEASURE
Ionizing radiation can be quantified using sev-
eral different units of measure.
43
The curie describes the amount of radioactivity
in a given amount of material such as a nuclear
core. A release of some of that material would be
measured as a certain number of curies. Subunits
are the microcurie one-millionth of a curie, and a
picocurie a trillionth of a curie.
The roentgen indicates the amount of X-rays or
gamma rays that will ionize a certain amount of
air.
A more general, but similar unit, the rod, is the
dose of any type of radiation (X-rays, alpha par-
ticles, etc.) that delivers a fixed amount of energy '
to some material (such as tissue, air, etc.).
The rem is a more useful measure of dose for
those concerned with health effects. It takes into
account the different biological damage done by
different kinds of radiation. One rad of alpha
radiation may result in a dose equivalent to 10 rem,
whereas one rad of X-rays to the same tissue could
result in a dose equivalent to only one rem. The
rem allows the health effects of radiation releases
to be estimated more easily and the health effects
of different releases to be compared.
Because the rem is a larger dose than normally
occurs in routine exposure to radiation, dose equiv-
alents are generally expressed in millirems
(mrem), or thousandths of a rem.
The rate at which exposure to radiation occurs
is expressed as the dose rate per hour. A person
receiving a dose of 100 mrem over a period of one
hour is receiving a dose rate of 100 mrem/hr. An-
other unit of dose rate is rads/hr. If a release in-
volves different types of radiation producing doses
of varying amounts of millirems per hour, the
total dose rate would be the sum of various dose
rates.
The sum of the individual doses received by
each member of a certain group or population
within a specific area is called the collective dose.
It is expressed in person-reins. A thousand people,
each exposed to one rem, would have a collective
dose of 1,000 person-rems.
Another measure is the cumulative dose. This is
the total dose an individual or group receives over
a certain period. An individual who is exposed to
a dose rate of one rem/hr for five hours will amass
a cumulative dose of 5 rems.
RADIATION MONITORING AT TMI
INADEQUACIES IN MONITORING
Because the monitoring of offsite releases in the
early stages of the accident was inadequate, it has
been difficult to determine the total amount of
radioactive material released, especially on the
first day. and to determine the exposure of the
surrounding population. About 50 percent of the
portable radiation survey instruments were in-
operable. 1 Only a limited number of fixed instru-
ments (3) were in place before the accident oc-
curred, and thev measured only total radiation ex-
posure, rather than dose rates. (4) Both factors
made it difficult for health phvsics personnel to
ascertain the rate of offsite radiation doses. (These
are important for projecting future doses of radia-
tion and for determining the need for evacuation or
other protective action.) Finally, offsite measure-
ments were not taken until about 8 :30 a.m.*
Some of these problems can be traced to inade-
quacies in the management of the health physics
program at TMI. The N"RC Special Inquiry
Group noted gaps in the radiation protection or-
ganization and stated that plant management and
operations staff regarded radiation protection as
a "necessarv evil." (5) The Special Investigation
found, as did the NRC Special Inquiry Group,
that on several occasions the utility transmitted
incorrect or misleading information on the radia-
tion levels measured by monitoring teams. 4
ESTIMATED RADIATION DOSES
For the above reasons, the exact exposure of the
population to radiation during the entire accident
is uncertain. Nevertheless, several groups have
developed estimates that are consistent.
The Ad Hoc Interagency Dose Assessment
Group, comprised of scientists from the Nuclear
Regulatory Commission. Environmental Protec-
tion Agency, Food and Drug Administration and
the Center for Disease Control, estimated that the
dose to the entire population within 50 miles of the
plant was between 1.600 and 5.300 person-rein,
1 Radiation imparts some of its energy to the medium with which it interacts. One rad equals 100 ergs of energy
delivered to one gram of material. The amount of radioactivity that produces one rad varies according to the type of
radiation.
2 The XRC found that only about half the portable radiation dose rate monitors (58 out of 107) were available. (1)
According to the report of the Task Group on Health Physics and Dosimetry of the President's Commission on Three
Mile Island, the high percentage of inoperable instruments could have contributed to difficulties in getting data during
the first several hours of the accident before the Radiological Assistance Program (RAP) teams began to arrive, ami
to difficulties in achieving good health physics techniques. (2)
3 See "The Accident at Three Mile Island : The First Day." p. 112.
1 See fn. 2 above and "The Accident at Three Mile Island : The First Day," pp. 132-133. Since the accident, the utility
has still had problems with its health physics program. See "Recovery at Three Mile Island," pp. 175-177.
44
depending on what assumptions were used in the
calculation. (6) The Dose Assessment Group
stated that its calculations were based on conserva-
tive assumptions which "introduced significant
overestimates of actual doses to the population." 5
(8) The Group also estimated the average dose to
an individual to have been 1.5 mrem. (9)
The Dose Assessment Group concluded the ef-
fects of offsite releases were minimal throughout
the accident. 8 (10)
The President's Commission generally agreed
with the figures derived by the Dose Assessment
Group and. like the Group, concluded that it was
possible to derive reliable estimates: ". . . these
deficiencies [related to measuring releases of
radiation] did not affect the Commission staffs
ability to estimate the radiation doses or health
effects resulting from the accident." (11)
The XRC Special Inquiry Group also concurred
in the dose estimates, concluding that the average
dose to an individual was about 1.4 mrem. (12) It
likewise said that ". . . although the monitoring
efforts could have been better .... the monitoring
of releases during the accident was adequate to en-
sure that the estimates of dose to the population
are adequate." (13)
A test by the Food and Drug Administration
of the U.S. Department of Health. Education and
"Welfare provided additional support for the
estimates. Scientists from the Bureau of Radio-
logical Health of that agency collected photo-
graphic film from stores near the site to ascertain
if it had been fogged by radiation and, if so. what
the levels of radiation had been. It did not find
abnormal or excessive fogging.' It concluded that
if the fogging had been produced solely by radia-
tion, the exposure levels would have been less than
5 mrem. 8 This finding is in line with other
c-timates. (15)
Based on these estimates, the average total dose
to an individual from the accident was about 1.4
to 1.5 mrem. Total dose rates were less than 6 mrem
per hour.
COMPARATIVE DOSES
The following examples of doses are provided
for purposes of comparison: (16)
Xo observable adverse health effects result
from a short-term dose to the entire body
of less than 25.000 millirem (mrem) (25
rems). Severe adverse health effects (radia-
tion sickness) are observable within two
hours for doses of 200.000-600.000 mrem
(200-600 rems). Immediate lethal effects
result from doses in excess of 1,000,000
mrem (1,000 rems).
The U.S. population receives an average of
about 100 mrem per year from natural
background radiation (e.g.. from the sun,
radiation from buildings, soil, ete.). Be-
cause Denver. Colorado, is at a high alti-
tude, and consequently less radiation is
filtered by the atmosphere, the rate is 193
mrem per year. In Harrisburg, Pa., near
TMI. background radiation is 116 mrem.
The U.S. population receives an average of
100 mrems per year from medical diagnoses.
A chest X-ray using good equipment pro-
duces 15 mrem.
The XRC standard for nuclear power plant
workers is a whole body dose of 3.000 milli-
rem, or 3 rem. every 3 months. The EPA
standard for individual exposure to radia-
tion from the uranium fuel cycle associated
with the operation of a nuclear plant for
one year is 25 mrem.
The average federally recommended limit
for exposure of the general population is
170 mrem ; for an individual it is 500 mrem
(1-5 rems).
1 The Group calculated total population dose using data collected from the utility's dosimeters in place before and
deployed during the accident, from XRC measurements and from DOE aerial surveys made during the accident. (7
' It calculated the amount of the releases both by extrapolating from releases measured at the boundary of the plant
site and by I lack-calculating on the basis of offsite measurements.
' Fogging also can be produced in other ways, such as by heat.
* Six rolls of Kodacolor 400 film, recommended by Kodak for this purpose, were collected from each of five sites
within a few miles of Three Mile Island. The film was analyzed for fog levels by the Bureau after processing by Kodak.
A batch of film purchased in Rockville. Maryland, was used as a control.
When both sets were developed, that from Rockville showed similar levels of fogging to that from the TMI area.
When compared with film of similar age stored in freezers at Kodak, these fog levels were found to be smaller than those
of the Rockville film. (14).
45
Chapter 6
Prior To The Accident
47
Three Mile Island under construction
48
Chapter 6
Prior To The Accident
INTRODUCTION
The Special Investigation explored the period
prior to the accident as part of its examination of
why a minor transient was able to escalate into a
major accident and why the responses of the util-
ity, the XRC and the State were inadequate. To
this end. the development of Unit 2, 1 the nature of
accidents at other facilities and the emergency
response planning of the three organizations were
reviewed.
THE EVOLUTION OF UNIT 2
The Special Investigation's review of the
design, construction and early operating experi-
ence of TMI-2 was instructive as to how decisions
about the plant were made and what types of
operational and other difficulties had occurred.
Some of the problems that emerged from this
review bore directly on the March 28, 1979 acci-
dent. Among the more important were a number
of deficiencies in control room design and instru-
mentation that control room personnel had identi-
fied and had asked to have changed. For example,
they requested that a direct indicator of the posi-
tion of the pilot-operated relief valve (PORV) be
installed in the control room. The utility installed
an indirect indicator which, on March 28. misled
the operators into thinking the valve had closed. In
fact, it had stuck open, allowing coolant to escape
the reactor vessel. 2 A further example was the
alarm system. In the event of a major accident,
hundreds of alarms would activate in the first few
minutes, far more than the control room personnel
could assimilate. In addition, because of the design
of the system, in the process of clearing an alarm
it was possible to acknowledge others that had
sounded but were not yet noticed. Although modifi-
cations were made to the alarm acknowledgement
system, they were insufficient according to the
control room personnel, who decided, before
March 28, 1979, not to acknowledge any alarms
in the first minutes of an accident. 3
Control room personnel had also come to dis-
count key indicators of abnormal conditions be-
cause of recurrent equipment malfunctions. The
safety-related emergency high pressure injection
system, designed to activate for losses of coolant,
was coming on for less severe problems. 4 One or
more of the valves on the pressurizer was leaking,
causing elevated temperatures in the lines leading
from it. 5 On the day of the accident, control room
personnel did not interpret actuation of high pres-
sure injection to mean a loss of coolant, nor did
they interpret the even higher valve temperatures
to mean the pilot -operated relief valve was stuck
open, allowing coolant to escape. 6
Various incidents that occurred during testing
and startup of the reactor during the years prior
to the accident would reoccur March 28. For ex-
ample, in 1977 steam became trapped in the hot-
legs and blocked the flow of coolant. The level
of coolant in the pressurizer went up, while pres-
sure in the primary system dropped an unusual
occurrence. On March 28, similar conditions oc-
1 It was beyond the resources of the Special Investigation to examine all facets of Unit 2's development. The AEC's
and NRC's involvement in licensing and inspection, and management's position on many of the issues raised since the
accident, could not Ue fully addressed. Information on early design and operating problems was derived principally from
control room personnel.
1 See "The Accident at Three Mile Island : The First Day," p. 94.
1 See p. 69. ' See p. 72. 'See pp. 71-72.
' See "The Accident at Three Mile Island : The First Day," pp. 96. 108.
49
curred, but control room personnel on duty ap-
parently were unaware of the early problems, did
not understand the conditions, and responded in
ways that contributed to a worsening of the loss
of coolant. 7
Training was an area of importance, given the
inability of plant personnel to diagnose the acci-
dent and their ineffective attempts to return the
plant to stable conditions for much of the first day.
The Special Investigation found major deficien-
cies in the utility's training program and the
NEC's oversight of training, as did other inves-
tigations. 8
Other aspects of the plant's pre-accident history
are less directly related to the events of March 28.
Management of the design and construction of the
facility was fragmented. For example. Met Ed,
the utility that ultimately operated the plant, had
limited involvement in decisionmaking until after
the plant was fullv constructed. 9 Several future
operators of TMI-2 said that control room person-
nel had little to do with evaluation of the final
design, particularly of the control room, prior to
startup operations, although they were responsible
for plant operations. 10
Economic considerations quite naturally in-
fluenced decisionmakinj?. When the plant was
transferred from the New Jersey site to Penn-
sylvania, General Public Utilities, the parent com-
pany, established a policy of minimum change.
Although Unit 2 met all NEC safety require-
ments, some desirable changes in the final design
of Unit 2 identified bv plant personnel during the
final construction and early operation of the, plant
were not made, in part for economic reasons, and
the weaknesses those changes would have corrected
contributed to the difficulties utility personnel had
in responding to the accident on March 28."
ACCIDENTS AT OTHER PLANTS
The problems at TMI were not entirely unique.
Prior to 1979, two other plants had experienced
accidents that were quite similar to the early stages
of TMI-2. Both were diagnosed in time to help
prevent the later, serious conditions experienced
at TMI. Information regarding these accidents
was not effectively disseminated industry-wide by
the NRG or by the vendors of affected systems. 12
EMERGENCY RESPONSE PLANNING
The responses of the utility, the NRC and the
State to the accident revealed that their emergency
response planning was seriously deficient. Prior to
the accident, there was no coordination among the
three providing for an integrated response. This
was especially apparent with respect to considera-
tion of protective action. 13 Responsibilities were
not carefully delineated, and inadequate means
were developed for communicating and assessing
information on the status of the reactor. 14 Federal
guidelines promulgated by the Environmental
Protection Agency were vague and gave insuffi-
cient guidance to 'the State and the utility with
regard to what information was germane in assess-
ing the status of the reactor, how that information
was to be collected and bv whom, who was to re-
ceive it, and how that information was to be used
as a basis for taking action to protect the public. 15
In addition, none of the three had adequate
technical or manpower resources available at the
outset of the accident. All experienced serious dif-
ficulties with both internal and external communi-
cations. 16 These problems were, in part, a result of
limited assumptions that were made as to the kinds
of accidents to be anticipated. 17
THE EVOLUTION OF UNIT 2
PLANNING
Three Mile Island Unit 2, constructed between
1969 and 1977, is located on an island in the Sus-
quehanna River. 10 miles southeast of Harrisburg,
Pennsylvania. The surrounding area is still pre-
dominately rural and agricultural, but recently
there has been substantial industrial development.
Within a five-mile radius of the plant, the popula-
tion numbers about 30,000-35,000; within 10 miles.
125.000-135,000; and within 20 miles, 750.000-
900,000. (1) The nearest town Goldsboro, popu-
lation around 550 is n/ 2 miles west of the
facility.
TMI-2 is owned jointly bv three operating com-
panies: Metropolitan Edison Company (Met
Ed 'I 50 percent : Pennsylvania Electric Company
(PENELEC) 25 percent; and Jersey Central
'See p. 65. 'See pp. 73-76.
" See pp. 84-86. " See pp. 84-86
"See pp. 51-58.
15 See pp. 85-86.
' See p. 58. " See pp. 59-60, 66-72.
18 See pp. 82-83 " See p. 83.
12 See pp. 76-78.
50
Power and Light Company (Jersey Central) 25
percent. (2) All are wholly owned subsidiaries of
General Public Utilities Corporation (GPU) , an
electric utility holding company headquartered in
Parsippany, Xew Jersey. (3) Under the license
issued bv the XRC. operation of TMI-2 and its sis-
ter plant. Three Mile Island Unit 1 (TMI-1), is
the responsibility of Met Ed.
Two aspects of TMI-2's early development are
significant. First, it was initially planned for an-
other site, and the decision to move it to Three Mile
Island only came after the preliminary design had
been completed. Second, management of the plant
in the early years was scattered among GPU and
its subsidiaries.
These factors affected the plant's ultimate
design and decisions about modifications requested
by TMI operations staff.
FRAGMENTED MANAGEMENT
In the early 1960's. GPU decided to build a sec-
ond nuclear plant at its Oyster Creek site in New
Jersey. Ownership was to be shared by the three
GPU subsidiaries: however, Jersey Central, re-
sponsible for Oyster Creek Unit-1, was to operate
it. (4)
While planning for the Oyster Creek 2 project
was getting underway. GPU decided to aggregate
the company's nuclear resources, thus eliminating
redundancy among the managerial and technical
staff of it? three subsidiary operating companies.
(5) Thus, in 1967, GPU established the Nuclear
Power Activities Group to serve as central coordi-
nator for the design and construction of all its
nuclear projects. (6) The parent company's plan
was that the. three subsidiaries would eventually
be responsible only for operating completed plants.
(In 1967, no GPU nuclear plant was operational.)
In 1968. a year after the Activities Group was
established, the operating companies still exer-
cised many of the functions the Activities Group
had been set up to take over. (7) Jersey Central,
for example, had solicited bids for the design and
construction of the new Oyster Creek plant and
had managed the preparation of the application
for a Construction Permit, submitted to the
Atomic Energy Commission on Anril 22. 1968. 18
(8) Met Ed was still managing the TMI-1 project.
Management after the Site Transfer
In December 1968, GPU decided to move the
Oyster Creek 2 plant to Three Mile Island. 19 (9)
Even though the plant was officially transferred
to Met Ed, Jersey Central continued to provide
technical support and had a major role in manag-
ing the project. (10) According to Tom Hendrick-
son of Burns and Roe, M part of GPU's rationale
for continuing involvement by Jersey Central was
to avoid delay in getting the plant on line. (11)
The change in location presented an opportu-
nity for either the Activities Group or Met Ed
to begin assuming principal responsibility, and
some organizational conflicts resulted. GPU's ob-
jective was to consolidate the engineering con-
struction management for all nuclear projects of
its subsidiaries under the Activities Group. (12)
On the other hand, Met Ed wanted to take
primary responsibility for TMI-2. and. in 1969,
it took administrative control of the TMI-2 proj-
ect for 18 months. (13)
As a result of the various maneuvers, there was
no continuity in the management oversight for the
project. (14)
In 1971, GPU initiated a second attempt to con-
solidate its nuclear programs. It abolished the
Activities Group and formed the General Public
Utilities Service Corporation (GPU Service Cor-
poration) as a subsidiary, (15) with basically the
same responsibilities. The new entity was to draw
on the resources of the three operating companies,
thus creating actual links among them. (16)
As GPU Service Corporation grew, it gradually
absorbed the design and development capabilities
of the operating companies. However, the process
was slow, and for seven years after it was estab-
lished, it shared oversight responsibility with the
GPU operating utilities. (17)
Management Problems Identified
By 1977, GPU commissioned a management
consulting firm, Booz. Allen and Hamilton, to con-
duct a managerial audit. Among the firm's conclu-
sions were :
An evaluation should be made of the au-
thority and responsibility of Met Ed func-
tional officers with respect to GPUSC
FGPU Service Corporation].
Policies that define the respective roles and
responsibilities of GPUSC and Met Ed in
the design and construction of new facili-
ties need to be reevaluated and clarified.
Communications between GPUSC and Met
Ed need to be strengthened in project-
related areas.
The effectiveness of present systems (main-
tenance) is reduced by their somewhat
limited application and use.
An approach and formal program should
be developed to improve the overall effec-
tiveness of ... maintenance ... at Met Ed.
There is a wide disparity in the quantity
and quality of plant operator procedure
documentation and training programs.
" See box on "The Nuclear Regulatory Commission Reactor Licensing Process," pp. 52-53.
1 The reasons for the move are discussed on p. 54.
" Burns and Roe, an architect-engineering firm, designed the Oyster Creek, later TMI-2, plant. See pp. 54-55.
51
NUCLEAR REGULATORY COMMISSION
In order to build and run a nuclear power plant at a particular site, a utility must obtain a Con-
struction Permit and then an Operating License from the NRC. 1 The NRC, like the Atomic Energy
Commission before it, 2 licenses, regulates and inspects the construction and operation of commercial
and other nonmilitary nuclear facilities.
Prior to construction, the utility submits to the NRC a Preliminary Safety Analysis Report and
an Environmental Report, which include information on safety design (in terms of construction,
equipment and systems); site characteristics; public health issues; personnel; management and
administration; emergency response plans; response to hypothetical accidents; environmental as-
pects; quality assurance; control of radiation effluents and wastes; and financial capability.
NRC staff reviews this material according to set procedures and criteria (these apply to sub-
sequent reviews as well). They may require additional material. The utility has to demonstrate that
it meets the requirements for licensing set forth in the NRC's regulations and has to justify any de-
parture from standards set by the NRC.
After examining the material, the NRC staff prepares a Safety Evaluation Report summarizing
its findings. The Advisory Committee on Reactor Safeguards, an independent body of experts estab-
lished by law, reviews the report and all background material. The full committee submits its find-
ings to the NRC Commissioners.
The NRC also provides a Draft Environmental Statement for analysis by Federal, State and
local agencies and the public. Comments are incorporated into a Final Environmental Statement.
Public hearings are required on all applications for a Construction Permit ; they are held before
an Atomic Safety and Licensing Board. If the Board issues favorable findings, the XRC will issue a
Construction Permit. Any decision by the Board may be appealed to and is reviewed by an Atomic
Safety and Licensing Appeal Board, and there is an opportunity for final review by the Commission.
The entire licensing procedure takes approximately 2,y 2 to 3 years, on the average, depending
on the design, the extent of the public hearings and the number of issues requiring further clarifica-
tion and justification.
At all stages, the utility is required to file amendments for any changes that would affect safety ;
these are subject to NRC approval.
When a utility has reached the point in plant construction where it is able to present final design
information and operating plans, it submits a Final Safety Analysis Report, a requirement for is-
suance of an Operating License. The NRC review procedure is similar to that for the Construction
Permit, except that a public hearing is not held unless requested (in accordance with the Commis-
sion's rules). In its final report, the utility must present additional material on the final design of the
facility, including data on the containment, the nuclear core and the waste handling system.
When the Final Safety Analysis Report is approved and any hearings are completed, the NRC
issues an Operating License with Technical Specifications for safety and environmental protection
measures and other operating criteria the utility must meet to ensure public health and safety.
1 See "The Nuclear Regulatory Commission Reactor Licensing Process," Appendix C, pp. 233ff, for more detail.
' The NRC inherited the AEC's regulatory authority when the AEC was abolished in 1975.
Formal guidelines and minimum standards
should be developed to help ensure con-
tinued safe, reliable power plant opera-
tions. (18)
In 1977, Robert Arnold, Vice President for Gen-
eration at Met Ed, was made Vice President for
Generation at GPU Service Corporation. He was
given responsibility for implementing some of the
Booz, Allen recommendations and for consolidat-
ing GPU's nuclear projects. (19) In addition, he
was to develop within GPU Service Corporation a
complete capability for basic design of nuclear
plants. Such a capability would permit GPU to
avoid contracting with architect-engineers. (20)
52
Arnold said of his mission :
[T]he issue was a very real one to us. It
was one that was emphasized by Herman
Dieckamp [President of Met Ed] when I
went into the job. of the need to couple
together the operating plant experience
with the plant design and to provide the
kind of technical review of what was hap-
pening at the plant that was necessary to
have the reliability of operation and
safety operation that was necessary. (21)
Arnold further commented :
We . . . established a procedure which
we had a great deal of difficulty getting
REACTOR LICENSING PROCESS
An Operating License can be issued even while some safety-related questions are unresolved. For
example, if it is determined that a question involves a generic issue, that is, it applies to a class or
type of plant, the issue may be left unresolved until a final generic solution is developed. 3 A license
cannot be issued, however, if the XRC staff determines that the question involves a safety factor
that should bar continued operation or should require licensing actions prior to completion of the
longer term review. (26) These same stipulations apply to non-generic safety issues.
Once a plant is in service, the utility must file with the NRC a Licensee Event Report (LER)
if it experiences an unusual incident (e.g. a transient). The XRC responds to these reports in sev-
eral ways. They are placed on a computer, and a listing is published each month. Each event is also
briefly summarized in a monthly publication, NUREG-0020, "Operating Unit Status Report."
XRC staff in the Commission's regional offices and, to a lesser extent, at headquarters also re-
view Licensee Event Reports for incidents of special safety, safeguards or environmental signifi-
cance and to ascertain whether they have generic implications. Licensees are informed of generic
concerns through Bulletins, to which they must respond in writing; through Circulars, to which
they need not respond ; or through Generic Letters or Orders.
The XRC licensing functions associated with construction and operation of nuclear reactors are
performed by the Office of Xuclear Reactor Regulation. There were four main divisions in the Office
which carried out this responsibility at the time of the accident at TMI.
The Division of Project Management was assigned management functions in connection with
reviews of reactor safety up to the time an Operating License was issued.
The Division of Systems Safety carried out the actual safety reviews connected with applica-
tions for Construction Permits and Operating Licenses.
The Division of Site Safety and Environmental Analysis reviewed all safety and environmental
aspects of reactor sites.
The Division of Operating Reactors took over from Project Management and Systems Safety
responsibility for reviewing proposed design and operational changes at operating reactors.
Thus, when an Operating License was issued, primary responsibility within XRR for a plant was
transferred from the Division of Project Management to the Division of Operating Reactors.
There sometimes was a delay in this transfer of responsibility if a safety issue was unresolved.
The Division of Operating Reactors did not accept responsibility for TMI-2 until August 1979, five
months after the accident, because there had been such unresolved questions. (27)
Plant construction and operations are monitored by the Office of Inspection and Enforcement
(ME). It consists of a headquarters group and five regional offices, charged with the development
of policies and the implementation of programs for inspection of licensees, applicants and their
contractors and suppliers to ascertain whether they are complying with XI?C regulations, rules,
orders and license conditions. Inspectors from this Office were monitoring TMI-2 during the period
immediately prior to the accident.
' An example of a generic problem found at operating plants was an error discovered in calculations relating to
the performance of piping systems during earthquakes, a problem that involved several plants.
executed reliably, so I would not want to
take too much credit for what it was, but
a policy was set out and it is indicative
of what we were putting into place as one
of the ways to address this problem. (22)
* * *
... I don't think we felt at any point
that the structure we had was inadequate
or inappropriate. We rather felt that
there were ways in which we wanted to
improve it as we kept building toward
the future. (23)
When the Operating License for TMI-2 was
issued by the XRC, on February 8, 1978, complete
responsibility for the operation and safety of the
nuclear plant devolved to Met Ed, although for a
time GPU was to retain responsibility for com-
pleting the project and for the schedule. (24)
TRANSFER TO THREE MILE ISLAND
By the end of 1968, a large portion of the facility
and equipment for Oyster Creek 2 had been pro-
cured. However, the AEC had not yet issued a
Construction Permit, and no major construction
work had been undertaken. (25)
53
In December 1968, there was a meeting involv-
ing : GPU and its subsidiaries ; Burns and Roe, the
architect-engineer; and the major contractors and
vendors for the facility. (28) Among the latter
was Babcock & Wilcox, which was to supply the
nuclear reactor. 21
At the meeting, GPU announced a decision to
transfer the plant to Three Mile Island. (30) Ac-
cording to Louis Roddis, then Director of the Nu-
clear Power Activities Group,
The problem was related to construction
labor difficulties in the central New Jersey
area at that time frame which were
basically resolved after the Colonial
Pipeline cases came to trial and were
settled. It was just a very unfavorable
labor climate to operate in. (31)
Burns and Roe's Hendrickson elaborated on
Roddis' explanation in an interview with Special
Investigation staff :
Late during the construction phase of
Oyster Creek Unit No. 1, the Oyster
Creek Plant, there were difficulties ex-
perienced by the utility, basically labor-
related difficulties in the Jersey com-
pany's area down there where it was being
built. They were not anything that we
were involved in except that a decision
was made by General Public Utilities to
relocate the unit from the Oyster Creek
site to the TMI island site. (32)
Prior to the meeting, GPU had commissioned a
comprehensive study of the factors involved in
the transfer. (33) A summary of the findings was
spelled out in November 1968 in a memorandum
Roddis sent to GPU President William G. Kuhns.
(34) The following are the key points made in the
memorandum : (35)
Subject :
Cost
Site problems at TMI _
Ocean discharge
Plant reliability
Construction schedule
Construction contractor-
Study conclusions :
Operating cost at TMI
would be less, with a
possible annual sav-
ings of $100,000 In
operating labor ex-
penses.
Manageable.
No problem at TMI.
No change.
A delay was anticipated
with the transfer.
The Architect Engineer
should be the same
for both plants at TMI.
DESIGN QUESTIONS
Transferring the plant involved some difficul-
ties. For example, dovetailing the construction of
the two TMI units was possible to only a limited
extent. 22 Although the basic reactor designs were
the same (Babcock & Wilcox pressurized water
reactors), overall plant designs differed. Much of
the equipment Burns and Roe had chosen for the
Oyster Creek plant was of different origin than
that for the TMI-1 plant, then under construction
by the architect-engineering firm of Gilbert As-
sociates, Inc. (37)
There were differences of opinion within both
GPU and Met Ed as to whether the two plants
should have different designs. (38) Both com-
panies had discussed with the other operating
companies the possibility of completely redesign-
ing TMI-2 to match TMI-1, (39) biit economic
and scheduling considerations ran counter to this
plan. (40)
Another issue was whether Burns and Roe
should remain the architect-engineer for TMI-2.
(41) Met Ed hnd a longstanding contractual
relationship with Gilbert Associates, Inc., because
of TMI-1 and previous power plant projects. (42)
Further, Roddis believed that Gilbert was "a bet-
ter design engineer." (43) He added,
31 The report of the Special Inquiry Group provides a concise description of the relationship among the various
organizations that play a part in the construction of nuclear plants:
Some of the first commercial nuclear plants were "turnkey" projects designed and built entirely by vendors
General Electric or Westinghouse for a fixed price and then turned over to utilities for operation. In this
country that pattern has changed. A utility now hires an architect-engineer firm like Bechtel, Stone &
Webster, or, in the case of TMI-2. Burns and Roe, to serve as "general contractor," design the overall layout
of the plant, and serves as the utility's technical advisor in buying a vendor's reactor system. ... It is typical
to have different companies involved in the construction and operations of the plant. (29)
22 There is a growing practice of preplanning multi-plant power stations. The theory is to design the plants identically,
then construct them in tandem. The typical scheme is to stagger the schedules of twin plants by t.vo or three years,
allowing the development stages of design, engineering and construction to be shifted sequentially from the lead plant to
the following plant, with substantial economic savings. Maintaining the same plant design also affords operational
conveniences. As .Tames Neely, GPU project manager for Oyster Creek in 1968, stated :
It is generally accepted practice that if you have two plants . . . duplicate on the site, you are in a much better
position from the standpoint of overall operability and maintainability than if you have two different plants on
the same site. (36)
Such a construction scheme was even less feasible at Oyster Creek because of differences in plant designs and sched-
ules. For example, Oyster Creek 1 was a boiling water reactor, while Unit 2 was to be a pressurized water reactor.
54
If we were building TMI-1 and TMI-2
as a paired plant at that location, I cer-
tainly would have one AE [architect-
engineer] for the whole job, and in the
time frame of 1966, whenever that deci-
sion was made by Metropolitan Edison to
choose, Gilbert, 'it [Gilbert] would have
been the one for both of them. (44)
Rocldis later explained his comments by saying,
"We didn't have that option. That was the point
I am trying to make." (45) He also noted:
[The TMI-2 design] was adequate. It was
in the licensing process [Construction
Permit] at an advanced stage. It was
being done by an architect-engineer that
was competent ... it was different than
Unit 1, but there was nothing that said
necessarily that any feature of it was bet-
ter or worse. (46)
A MINIMUM CHANGE POLICY
At the same meeting where the transfer was
announced, GPU stated that final design and con-
struction of TMI-2 were to involve "minimum-
change." (47) This objective was reflected in
Burns and Roe's minutes of the meeting :
It is a requirement that the minimum pos-
sible disturbance be made to the existing
design, so as not to detract from the
schedule. A design will be used, even
though not optimum, provided it is ade-
quate and can save time. (48)
According to Neely, "The overall decision to
move the plant with the minimum changes was
based on economic considerations." (49)
In accordance with the GPU directive, Burns
and Roe remained the architect-engineer for TMI-
2, and construction proceeded with significant dif-
ferences in the design of the two TMI plants. 23
(50)
Changes Still Required
The transfer necessitated some major changes
in the original Oyster Creek design. For example :
1. The heat rejection system, which releases un-
usable waste heat from the plant to the environ-
ment, was converted from an open circulation 24
Major buildings of Unit 2, Three Mile Island Nulear Station, including the Epicor-If
M For example, the turbine generator and condensate polishers for the two plants were supplied by different
manufacturers ; operationally. Unit 2 was designed for a higher level of power generation than Unit 1. The control room
designs were also different. ( The TMI-2 control room is discussed in detail on p. 61. )
-' In open circulation, cooling water is continuously drawn from a body of water (e.g., the Atlantic Ocean near the
Oyster Creek site) and then discharged back to it at a slightly higher temperature because of the waste heat from the
plant. In closed circulation, cooling water heated by waste heat is pumped through a recirculation loop and into the
cooling towers. The heat is transferred to air drawn through the structure of the towers. Thus both methods discharge
heat to the environment, but open circulation systems heat the water, whereas closed systems heat the air.
55
to the closed circulation design of TMI-1, which
was using the multi-story cooling towers (51) that
have become a familiar feature of nuclear and
other power plants.
2. Major structures in both TMI units, includ-
ing the containments that house the reactors, were
reinforced to withstand the collision of a jet air-
liner landing or taking off at the nearby Harris-
burg airport. Since such a crash might involve in-
stantaneous explosion and burning of jet fuel,
safety-related design changes also had to be made
in the TMI corridors and ventilation system. (52)
3. The foundations of the plant were modified
to accommodate the difference in terrain between
Oyster Creek and TMI. The TMI site was com-
posed of bedrock, whereas Oyster Creek had a
more penetrable composition of sand and gravel.
(53)
The AEC required that the utility provide a
safety assessment of the proposed modifications. It
was submitted, and on November 4, 1969, the AEC
granted Met Ed a Construction Permit for TMI-2.
THE TMI-2 CONTROL ROOM
The control room is the operational center of a
nuclear power plant. From it, the operator, using
his training and experience and assisted by written
procedures, assesses the status of the plant and con-
trols its operation. In the event of an accident, it
is primarily in the control room that operators
diagnose the problem and take corrective actions
to bring the plant to a stable condition.
The Special Investigation's review of the evolu-
tion of the control room revealed that several de-
ficiencies had surfaced during the construction and
testing phases, as well as during two minor acci-
dents that occurred after the plant went critical
in 1978. Operators and supervisors had requested
that management modify manv of the trouble-
some features, but at the time of the March 28 ac-
cident, a .number remained unchanged or had been
changed unsatisfactorily in the operators' opinion.
(54)
PLANNING OF THE CONTROL ROOM
Human Factors Engineering
To achieve a workable design for a control room,
design engineers, working within the broad
framework of NRC requirements, apply their ex-
pertise and experience with instrumentation and
controls. (55) An important consideration in de-
veloping a control room is "human factors," and
a control room should be designed in accordance
with human factors engineering practices. 25
NRC Requirements
Design of the TMI control room had begun in
1968. At the time, the NRC regulations had only
a few general requirements :
A control room shall be provided from
which actions can be taken to operate the
nuclear power unit safely under normal
conditions and to maintain it in a safe
condition under accident conditions, in-
cluding loss-of-coolant accidents. Ade-
quate radiation protection shall be pro-
vided to permit access and occupancy of
the control room under accident condi-
tions without personnel receiving radia-
tion exposures in excess of 5 rem whole
body, or its equivalent to any part of the
body, for the duration of the accident.
(56)
The NRC's guidelines did not explicitly cover
human factors aspects of a control room. (57)
Salvatore Gottilla, the lead instrument engineer
at Burns and Roe in 1969, commented on the
NRC's regulatory requirements :
Let me say briefly that there were no
regulatory guides or standards that dic-
tated the design of control rooms. There
were a number of regulatory guides
which had reouirements which imnacted
on the control room design . . . for ex-
ample, there is a commonlv uspd standard
in the industry: it's an IEEE standard
279. which Fwasl enforced as a require-
ment . . . for the design of safety shut-
down svstems. This had some require-
ments for the kind of equipment we use.
the wav we specify it, reauirements for
redundancy, requirements for pegging, et
cetera, et cetera, all of which, to some ex-
tent, impacted on the design of the control
room. (58)
CONTROL ROOM DECISIONMAKING
The TMI-2 control room was desi<rned primar-
ily by Burns and Roe. with input from the GPU
Service Corporation. Met Ed engineering person-
nel participated at times, corresponding indirectly
with Burns and Roe through the Service Corpora-
tion and occasionallv directly with Burns and Roe
through letters and memos. (59) On many occa-
"' The term "human factors" in the context of nuclear power plant design refers to the physical and psychological
needs and capabilities of plant personnel. Accounting for these needs and capabilities in the design nnd operation of
machines is called human factors engineering. Examples of human factors that became important at TMI on the day of
the accident were the ability of individuals to recognize, assess and respond to a barrage of Information and signals, and
their ability to ^vork in protective garments, such as respirators.
56
sions, representatives from GPU Service Corpo-
ration. Met Ed and Burns and Roe all participated
in conferences related to the project. (60) More-
over. Burns and Roe had contracted with many
other vendors, among them Babcock & Wilcox,
to supply instruments, controls and, in some cases,
entire prefabricated control panels. (61)
Thus, there was a considerable number of con-
tributors to the design and development of the
control room. (62) and differences of opinion oc-
curred. For example, in 1968. when the plant was
still planned for Oyster Creek, Babcock & Wilcox
had sent Burns and Roe some drawings of the
model control room at the B&W nuclear reactor
simulator facilitv (63) in Lynchburg. Virdnia.
About a month later. B&W proposed that Burns
and Roe use the design of the simulator control
room for the Oyster Creek 2 plant. (64) B&W
stated that to do so would be particularly advan-
tageous if plant operators were to train on the
simulator at the B&W facility. 26 (65)
Ed Gahan of Burns and Roe. the supervising
instrument engineer responsible for the design cri-
teria and review of the Oyster Creek, later TMT-2.
control room, advised against the proposal. (66)
In his opinion. B&W had built the simulator to
support their engineering design work, and it
lacked manv of the proper warning systems, indi-
cators and displays necessary in commercial opera-
tions to account for the human element. (67)
In a memo to Gottilla dated December 27. 1968,
Gahan expressed his thoughts on the B&W pro-
posal :
B&W had not explained how the sim-
ulator would be available for training.
Items found on actual control room
panels, such as annunciators, were not
present on the simulator panels.
Instruments and controls on the B&W
design were not of the "heavy duty
type consistent with power plant design
practice." (68)
Eventuallv. Gahan desipned the Ovster Creek
Unit 2 control room. (69) Babcock & Wilcox was
commissioned to supply three of its modular
panels.
By the time the decision was made to relocate
the Oyster Creek plant in 1968. the preliminary
work on control room design had been completed.
(70) In fart. Burns and Roe had already besrun to
tailor specific details around requests by the Oyster
Creek personnel who were to operate the plant.
(71) The design and engineering work slowed
while the utility management deliberated the pol-
icy and strategy of the transfer, and detailed de-
sign efforts were deferred while major changes in
plant systems were considered. (72)
One or Two Designs at TMI
As a result of the transfer, utility management
had had to choose between two opposing concepts
regarding design of the control room : (73)
1. To retain as much of the original Oyster
Creek design as the required plant system
changes would allow, or
2. To redesign the control room for Unit 2
to match the design of the TMI-1 control
room.
Those who supported redesigning the control
room argued that cross-licensing 27 of operators
would be simplified if the two control rooms were
identical. (74) Those opposed to redesign argued
that to have similar control rooms for plants with
different physical characteristics could confuse
cross-licensed operators. (75) Implicit in the latter
argument was that some aspects of the design of
the two plants at TMI would be different, in ac-
cordance with GPU ? s minimum-change objective.
In January 1969, Burns and Roe received a let-
ter from J. Bartman of Met Ed Operations, re-
questing that the TMT-2 control room be
redesigned to match TMI-1. (76) Gottilla con-
sulted with representatives of Jersey Central, who
advised that Bartman's request be ignored. (77)
Bartman persisted, and he had the support of
others at Met Ed. At a Burns and Roe conference
in March 1969, he again made his request. (78)
Redesigning the control room still conflicted
with GPU's minimum-change objective. (79) A
debate ensued over the cross-licensing issue, and a
call was made to the AEC. The agency confirmed
that similarity in control rooms was not a manda-
tory criterion for cross-licensing. (80) Those at
the conference accepted GPU's objective.
At the conclusion of the meeting. GPU stated
that
. . . [it] would have the final word on con-
trol room design changes, and that Burns
and Roe should accept no proposed
changes from Met Ed without prior ap-
proval of either GPU or JCPL [Jersey
Central]. (81)
PROBLEMS WITH THE DESIGN
Burns and Roe did encounter some problems
with the control room's design. For example, in
1971. its engineers discovered that the controls for
the feedwater system had accidentally been di-
vided between two of the main console panels and
were located 22 feet apart. (82) One of the two
M Prior to 1070 there was only one simulator in operation for a B&W-type reactor. Therefore, the B&W facility would
likely have l>een used by Met Ed no matter what control room design was nsed.
37 The XRC may license an operator to run more than one plant, a procedure known as cross-licensing. At Till only
senior reactor operators in superyisory positions became cross-licensed for both plants.
57
5t-OS8 0-80-5
panels had been supplied by Burns and Roe, the
other by B&W. After Burns and Roe informed
GPU of the problem, a series of design change
proposals was made and jointly studied by Burns
and Roe and GPU. (83)
Some of the proposals did not consider chang-
ing the position of individual instruments and
controls on the panels. Instead, the common sug-
gestion was to reposition several of the 17 major
panels in the control room. These preliminary at-
tempts proved impractical, for when the panels
were rearranged to regroup controls for one sys-
tem, those of another would become separated.
The designers finally became convinced there
was no simple solution. They resolved the problem
of the separated feedwater controls both by re-
arranging the positions of the control panels and
by redesigning several control panel layouts. (84)
REVIEW OF THE CONTROL ROOM
According to Bill Zewe, the Station Shift Su-
pervisor on duty at the beginning of the March
28, 1979 accident, in late 1973 and early 1974, the
shift foremen and shift supervisors who were to
work at TMI-2 were able to review and comment
on the design plans for the control room. Zewe said
his review was limited by the advanced state of
planning :
Certain little features that had not yet
been finalized, I had the time to comment
on those, but most of the engineering ef-
fort has already been completed by that
time, and had been pretty well set. (85)
Edward Frederick and Craig Faust, two opera-
tors also present in the control room at the start
of the accident, were comparatively junior operat-
ing personnel at this time. They said they were
not asked to review or comment on the design of
the control room at all during this phase. (86)
Design of the control room was substantially
complete in 1975.
Special Investigation staff asked Hendrickson
of Burns and Roe whether his company had ever
reviewed the control room in terms of operator
response to the accidents postulated in the Final
Safety Analysis Report submitted to the AEC.
He noted that since they did not have operating
procedures for Unit 2, 28 the designers had had no
opportunity to perform a "task analysis" 29 of the
control room and to modify the design based on
the findings. In his words :
Well, I think it's really very simple. We
did not do that [conduct the review] , and
we could not have done it because we did
not have the operating procedures.
Those were a matter that Metropolitan
Edison developed. And the only way it
[perform a task analysis] can be done is
to review the procedures against the con-
trol room. (87)
DEVELOPING PLANT PROCEDURES
James Floyd, Operations Supervisor for TMI-
2, was one of the people responsible for developing
operating and emergency procedures for TMI-2. 36
Special Investigation staff asked him whether sub-
sequent testing of the procedures helped opera-
tions staff identify problems with the design of the
control room.
Floyd responded :
Yes. The classic example, of course, is the
high-pressure injection flow meters. From
where you're controlling the valves, you
can't read the meters that you're trying
to control. It was identified as soon as the
procedure [was] delivered to the control
room. (89)
Floyd commented on the response of operations
staff to this particular problem :
[A solution] of course, would have been
to try to move the meters down to where
you could see them. But that involves the
ES system, engineering safety features
system . . . which have to have mechanical
as well as electrical separation. It would
have involved fire barriers and the whole
gamut of things involved, and it was
probably just easier to let the meters
[stay] where they were and take the three
steps if you needed to. (90)
OPERATORS IDENTIFY PROBLEMS
In 1976, while the control room was still under
construction but after the various panels had been
28 Operating procedures provide directions to plant operators controlling, monitoring and responding to the mechanical
systems of the plant in a normal state. There are also procedures for emergency and abnormal conditions.
M A task analysis is a review to determine the specific actions required of people performing a given function, for
example, monitoring or controlling machinery.
30 In general, procedures for TMT-2 were prepared in several stages. In the first, the procedures already in existence
for TMI-1 or original drafts by Met Ed, B&W, Burns and Roe or NTJS, a Met Ed consultant, were modified. Then they
were reviewed toy the Met Ed procedure writing group, after which they went to thp PORC (Plant Operations Review
Committee) for review and a first approval. (The Review Committee is a group of Met Ed operators and engineers who
advise plant management on reactor and radioactive waste safety.) The procedures were then "red-lined" and sent back
to the Review Committee for review and approval. ("Red-lining" refers to a process by which procedures are tested and
corrected.) The procedures then were sent to the Generation Review Committee for final review and approval. (88) The
NRC did not systematically review all procedures, and its approval was not required.
58
assembled, TMI-2 operators were able to familiar-
ize themselves with the room. (91) In 1977, testing
began (a phase that continued into 1978).
Throughout this period, operations staff ex-
pressed concern about certain features. (92) They
used several formal means for commenting on and
requesting changes in the control room design, in-
cluding the Field Change Request Form, the GPU
Startup Problem Report and the GPU Field
Questionnaire. (93) Reouests for design changes
were forwarded to GPU management for review
and possible implementation. (94)
GPU did not make all the requested modifica-
tions. Some it made only in response to actual dif-
ficulties. For example, Frederick said that with
respect to one problem, operators had asked re-
peatedly for a position indicator for the valves in
the feedwater system. (95) An indicator was not
installed until after a minor accident on April 23,
1978, which involved excess feedwater going
through a feedwater valve that had closed more
slowly than expected. 31 (96)
The Alarm System
Frederick pointed out another problem the
tremendous number of alarms in the control room :
When the operators first went over there
and started examining the control panels
as they were being built, we were im-
pressed right away with the number of
alarms (97)
He added, "that was a comment that we had from
the beginning, that the alarm system seemed rather
extensive." (98)
Alarm Acknowledgement
In addition to the number of alarms, operators
from the beginning expressed concern about the
system used to acknowledge and clear the alarms
in the control room. (99)
When an alarm activates in the TMI-2 control
room, one of about 1.200 2" x 3" annunciator win-
dows begin flashing brightly and a loud horn
sounds. (100) The operator acknowledges the
alarm from a control button on the main console,
causing the horn to stop and the alarm window
light to cease flashing and remain lit.
As soon as the cause of the alarm is taken care
of. the horn sounds and the alarm window light
begins flashing again, but more dimly than origi-
nallv. This is known as the "ring-back*' feature
of the alarm system. (101) The operator, by de-
pressing the same button on the console, can then
clear the alarm, silencing the horn and turning
the alarm window light completely off.
Conceptually, the ring-back feature of the alarm
svstem is a useful analytical tool for the operators.
However, with only a single button botli to ac-
knowledge and to clear lighted alarms, (102) op-
erators cannot acknowledge and clear them
independently. Without realizing it, they could
inadvertently acknowledge new alarms coming
into the control room while clearing previously
acknowledged alarms in the ring-back mode.
The TMI-1 plant, in contrast, had separate
buttons for acknowledging and clearing the
alarms. (103)
By the spring of 1978, after the hardware for
the TMI-2 control room was already purchased,
the operations staff requested that GPU redesign
the acknowledgement system for the alarms to
match that of Unit 1. (104) Floyd recalled GPU
management's response :
. . . the hardware that was [already]
purchased would not allow that kind of
acknowledging system . . . And it was
my understanding that the alarms in
Unit 2 could not be made to respond that
way without a tremendous additional
expense. . . . (105)
Additional Instrumentation
Operations staff also asked for additional in-
struments. Given the advanced stage of the
control room, the placement of additional instru-
mentation was makeshift. Floyd recalled two
specific cases. One related to the Liquid Waste
Disposal Panel (known as 8A) :
We added the panel 8A. which has come
under criticism; the one that has the re-
actor coolant drain tank instrumentation
[temperature and pressure] on it. To get
the indication into the control room, that
was the only spot that was available. So,
it was added at a back panel ; and hence,
out of the line of sight of the operator.
Some people may consider that a major
change. It was an addition and not added
in the proper location. (106)
This particular panel was a factor in the acci-
dent on March 28, 1979. Abnormal conditions in
the reactor coolant drain tank can be a sign of
a loss of coolant through a leaking or stuck-open
relief valve on the pressurizer. a situation that
occurred at TMI-2 that day. Since the indicators
of conditions in the tank were out of sight of the
main control room console and since there were
no strip chart recorders for these conditions, con-
trol room personnel found it difficult to see changes
and to track them over time. 32
The second case cited by Floyd related to the
location of some of the alarms on the panels.
When the sections of the panels containing the
alarm windows were designed, the need for addi-
tional alarms was anticipated, and excess win-
* This incident is described on pp. 66-70.
E See "The Accident at Three Mile Island : The First Day," pp. 100-101.
dows were provided. Twenty percent of the
windows at TMI-2 originally had not been desig-
nated for particular alarms. As time went on,
these windows were used for new alarms. Eventu-
ally it became hard to find a free window near the
appropriate control. Floyd commented :
. . . you tried to locate them in the area
where they were most useful to you for
the components controlled or alarming.
Sometimes this was not possible; there-
fore, you had to go clear to the other side
of the control room to find a planning
window to light. (107)
In trying to retrofit and redesign the control
room in the final stages of its development (see
box on the TMI-2 control room for details on the
final design) , Floyd stated that he
. . . wasn't free to move all the controls
around to get them the way I wanted
them, necessarily. So, it was a limited
choice that I exercised. (108)
According to Roddis, the TMI-2 control room
features generally did not compare favorably to
those of TMI-1 :
Well, it [TMI-1] has the feel in the plant
of having been laid out with somewhat
more consideration for the operator. For
instance, I was looking, when I was out
there a few weeks ago, at the purification
system, the water cleanup system, the con-
trol panel is much more thoughtfully laid
out, and the valve locations are near the
things you are trying to control. The same
unit in [TMI-2] is put together with
much less thought to the operator being
able to perform his functions easily. . . .
(109)
HUMAN FACTORS ENGINEERING
Many of the deficiencies in control room design
that TMI personnel had identified related to
human factors engineering. During the 1970's, both
industry and the Government began studying this
area of design. Up to that time, their principal
focus had been on designing safe mechanical
equipment and systems. The NRC and other
groups then became increasingly interested in the
potential for human error and saw a need to assess
its relation to the reliability and safety of nuclear
reactors.
The various studies revealed that insufficient
attention had been paid to human factors in de-
signing nuclear power plants and that there was
potential for human error attributable to poor
design.
One of the earliest studies was performed in
1972 by Dr. Alan Swain of Sandia Laboratories.
At the request of the AEC, he visited the Dresden
Nuclear Power Facility in Illinois. After his re-
view, he prepared a memo in which he discussed
some of the major departures from standard hu-
man factors engineering practices that he had
observed. (117) They included : (118)
1. The control rooms for Unit 2 and Unit 3
were mirror images of each other ;
2. A large number of displays and controls
were not grouped functionally; and
3. There was a constant barrage of alarms,
even under normal conditions.
The control room at Three Mile Island, Unit 2
60
THE TMI-2 CONTROL ROOM
As it was finally constructed, the control room looked as follows :
Back Panete Vertical Panels
Back
Computer
And
- 5 -' "6"
Adapted from Metropolitan Edison Diagram
The innermost I* consists of console-type panels, the tops of which are about chest high. Infor-
mation displays and control equipment used frequently during operations, the start-up controls
and the computer panel and protective equipment needed quickly in emergencies are all mounted
on these consoles. Included are the indicators and controls for the reactor power output, steam gen-
erators and turbine generator, reactor coolant make-up and purification system, safety features actu-
ation system and condensate and feedwater systems. (110)
Behind the consoles, and separated from them by a walkway, are the vertical panels. They stand
approximately seven feet high and contain the radiation monitoring equipment indicators, the indi-
cators and controls for the containment isolation valves, the individual control rod position in-
dicators, the status of engineered safety features, recorders for the temperatures of all major
equipment and of the primary system, and fire indicators. These panels also contain the annunci-
ator lights that are part of the alarm system. (Ill)
Behind the vertical panels are two back panels, out of the line of sight of the main console. They
contain the indicators for the heating and ventilating system, the sump pumps and the liquid waste
disposal system which include indicators for the temperature, pressure and level of the reactor cool-
ant drain tank. (112)
The alarm system in the control room consists of both visual and audible warning devices that
alert the operator if any system is approaching unsafe conditions. (113) The annunciator boards lo-
cated near the top of the vertical panels constitute the visual portion of the alarm system. They are
divided into approximately 50 individual boxes or windows, each a few inches on a side. Each win-
dow bears the name of a system or component. A computer with printout capabilities monitors plant
performance and alarms. It is also used for logging data. (114)
In a typical control room, the most common controls are selector switches, pushbuttons, rotary
knobs, thumbwheels, levers, toggle switches, and switch lights. (115) The most common indicators
are lights and meters. Strip chart recorders are also used extensively in control rooms to record
trends in given parameters over time. There are two basic types of chart recorders those that pro-
vide numerical printouts and multi-pen recorders that draw trend lines. The recorders help the
operator monitor the systems. (116)
A large number of controls, instruments, and alarms does not necessarily imply a better or worse
design. More important to ease of operation is the arrangement and display of control room devices.
61
Dr. Swain concluded that there had been no
formal or systematic consideration of human
factors technology in the design of the plant. (119)
In an interview with Special Investigation staff,
Dr. Swain said that, presumably as a result of this
work, the AEC decided to look more closely at
human reliability in the operation of nuclear
power plants. (120) The agency asked him to
participate in the preparation of WASH-1400,
the Reactor Safety Study, and he was one of the
primary contributors to the Human Reliability
section of that report, issued in 1975. (121)
The February 1974 edition of IEEE Transac-
tions on Nuclear Science contained an article en-
titled "Control Room Standardization: A Safety
Goal." In it, Dr. Stephen H. Hanauer, then Di-
rector, Office of Technical Advisor, Regulation,
AEC, raised two major concerns: (122)
1. In an emergency, a reactor operator is
relied on to perform important safety func-
tions for which he has been trained only on
a simulator. Therefore, the designs of simu-
lators and actual control rooms should be
similar. He recommended that since the num-
ber of simulators is small, the number of dif-
ferent control room designs should be small.
Hanauer called for the industry to standard-
ize its control room designs.
2. Control room designs were not optimal
in terms of safe reactor operations. Hanauer
suggested that some of the human factors en-
gineering used in the space program be ap-
plied to nuclear plants.
On March 13, 1975, Dr. Hanauer sent a memo
to NRC Commissioner Gilinsky, with copies to
Chairman Anders, Commissioners Kennedy, Ma-
son and Rowden, and other senior NRC staff.
The subject was "Important Technical Reactor
Safety Issues Facing the Commission Now or in
the Near Future." One focus was human per-
formance. (123)
According to Dr. Hanauer, the Commission did
not follow up on either document. (124)
Earlier in 1974. the AEC contracted with San-
dia Laboratories to do a human factors analysis
of a typical nuclear power plant to identify hu-
man factors problems and their effects on operator
reliability. Dr. Swain conducted the studv and in
October 1975 issued a report entitled "Prelimi-
nary Human Factors Analysis of Zion Nuclear
Power Plant." (125) His major findings were:
1. Standard human factors techniques
could be used to identify inadequacies in the
design of equipment, in the provisions for
training and practice, and in operating
procedures.
2. Control room design deviates in many
ways from accepted human factors engineer-
ing standards and increases the probability
of human error. Swain cited, as examples,
the poor layout of controls and displays, the
excessive number of annunciators (alarms),
misleading and inadequate labels on controls
and displays, and a confusing use of color to
indicate the status of equipment.
He concluded that :
1. Some relatively minor and inexpensive re-
designing of equipment, more emergency re-
sponse drills and changes in the format and
content of written procedures could improve
human reliability.
2. Valuable data on human performance
could be collected for detailed quantitative
human reliability analysis studies. (126)
One of the report's principal recommendations
was that industry-wide standards be developed on
the application of human factors engineering to
equipment, written procedures, operating meth-
ods, and onsite training and practice for nuclear
power plants. (127)
Based on this report, the NRC's Human Engi-
neering Research Review Group 33 recommended
that a Regulatory Guide be prepared on control
room design. According to William Farmer,
Chairman of the Group, there was some agree-
ment within the NRC that such a guide was
needed, (128) but it was never developed.
The NRC sponsored another study to provide
data for a standard being developed by the Amer-
ican National Standards Institute on Criteria for
Safety-Related Operator Actions (ANSI 660).
The objective was to determine the amount of
time an operator needs to respond to a situation
and to adjust controls or take corrective actions.
The study also was to assist the Office of Nuclear
Reactor Regulation, NRC, in deciding when to re-
quire automatic responses by plant equipment and
when an operator could be responsible for taking
corrective action.
The preliminary data supported NRR's gen-
eral standard that if response were required in less
than 10 minutes, it should be automatic. (129)
In 1075, under contract to the XRC. The Aero-
space Corporation began to assess the effect of con-
trol room design on operator performance during
stressful conditions. (130) The study also briefly
addressed the impact of operator training and
emergency procedures on operator performance.
Seven nuclear facilities were visited in the course
of the study.
In February 1977, The Aerospace Corporation
issued its report, "Human Engineering of
Nuclear Power Plant Control Rooms and Its Ef-
33 An interoffice group of staff interested in human factors. See "Nuclear Regulatory Commission Organization,"
Appendix B, pp. 227ff.
62
fects on Operator Performance." (131) It identi-
fied several weaknesses in control room design, such
as: (132)
1. The layout of the controls and instru-
ments combined with the number of actions
required of an operator could lead to serious
errors under accident conditions.
2. The color systems used to indicate the
status of equipment were confusing under
both normal and accident conditions.
3. Control panels that contain row upon row
of identically shaped push buttons and/or
switch handles may lead to operator error.
The report questioned the usefulness of emer-
gency procedures during stressful situations and
emphasized the value of simulator training to pre-
pare for emergencies. It, too, mentioned the limited
number of control room simulators available for
training and the fact that simulators were dissim-
ilar from actual control rooms. (133)
Major recommendations were: (134)
1. The XRC should develop a Regulatory
Guide to provide direction for utilities in hu-
man factors engineering as applicable to
control rooms and to encourage the use of ad-
vanced concepts for controls and displays.
2. Useful data should be collected on the
nature and frequency of operator errors as
part of an assessment of the effectiveness of
different control room designs.
3. A study should be conducted to determine
whether available simulators are capable of
providing operators with the training needed
to minimize errors under conditions of severe
stress. The study also should evaluate the ef-
fectiveness of training on a simulator that does
not realistically correspond to the actual lay-
out of the control room.
The report endorsed the use of mimic-type flow
diagrams 34 on control panels to help operators un-
derstand the relationship of key system compo-
nents, as WP]] as the trend toward the use of
cathode ray tube displays 35 and computers in the
control room. (135)
The report was widely distributed within the
XRC in 1977. (136) However, according to
Thomas Ippolito. Branch Chief in the Office of
Xuclear Reactor Regulation, because of other pri-
orities and limited resources, a Regulatory Guide
was not developed, nor were follow-up studies
conducted. (137) However. Sandia Laboratories,
in an XRC-sponsored study, did start collecting
data on operator error rates. 3 " (138)
In March 1977. the Electric Power Research
Institute (EPRI) 37 published a report on human
factors. It was entitled "Human Factors Review of
Nuclear Power Plant Control Room Design,"
(139) and was the result of a study carried out by
the Lockheed Missiles and Space Company, Inc.
under contract with EPRI. Lockheed had con-
ducted a survey at five representative control
rooms at operating nuclear plants. The report cited
significant problems relating to human factors,
such as: (140)
Certain instruments and controls could not
be read easily because they were too high or
too low, the lighting was poor and the labels
were too small.
The absence of standards for color codes,
control dimensions, label descriptions and
abbreviations promoted confusion.
The control panel layouts lacked functional
grouping of related controls and alarms.
The layout of some control rooms hampered
the operator's ability to respond to an inci-
dent.
The large number of alarms distracted the
operator from identifying and resolving a
problem.
Written emergency procedures were in some
cases incompatible with control room
design.
The report stated that the most convincing evi-
dence of deficiencies in control room design were
the design modifications that the operators had
introduced to improve their response in emer-
gencies. (141)
The report had wide distribution within the
NRC, and the NRC's Human Engineering Re-
search Review Group held meetings to discuss it.
(142) However, the NRC took no specific action.
(143)
The various reports stressed common themes.
They warned that inadequate attention was being
paid to human factors engineering at nuclear
power plants. They cautioned that the risk of hu-
man error was increased by design features which
were incompatible with the needs and capabilities
of plant personnel. They repeatedly urged that de-
sign standards and a Regulatory Guide be devel-
oped in this area.
In spite of these findings, the KRC did not issue
the Guide. It gave as the primary reasons a heavy
workload and limited budget for technical assist-
ance. (144) Human factors engineering was as-
signed a low priority, and other regulatory mat-
ters took precedence. (145)
The accident at Three Mile Island was to bear
out many of the predictions made in the reports.
31 A sketch of the system is superimposed on the control panel. Controls and indicators are placed on the panel at the
positions on the sketch which correspond to their system function.
35 A cathode ray tube display uses a tube similar to the picture tube in a television set.
"This addresses the probability that an operator's action or task will not be completed successfully within the
required time.
31 EPRI is a research institute supported by electrical utility companies.
63
TMI-2'S DESIGN IN RETROSPECT
Two senior officials of GPU and Met Ed ac-
knowledged that Met Ed operations personnel had
had limited participation in the design of the
plant at the time construction was nearly com-
plete. GPU President Herman Dieckamp stated :
I think Met Ed people did, to some de-
gree, participate in the design reviews,
even though I am sure that was not as
extensive as ... the operating people say
they should have had. ( 146)
Met Ed President Walter Creitz stated that:
There were opportunities for general in-
put available during the period of con-
struction, and yet I must admit that some-
times a person might observe a proposed
change, and it could be too late ; maybe it
wasn't identified on the drawing. (147)
* * *
... I remember walking through the
plant with Gary Miller and/or Jack Her-
bein, and various things might have been
pointed out, like the valve example ; this
shouldn't be here, it should be here, or we
should have done this, or we should have
done that. I guess you learn from experi-
ence. Perhaps, it is just that man is not
capable of putting down on paper the
ultimate in what he would like to build.
(148)
EARLY OPERATING EXPERIENCE
TMI-2 PLANT TESTING
The TMI-2 reactor went critical on March 28,
1978, one year to the day prior to the accident. Be-
fore it went critical, the utility spent about a year
conducting a number of tests to ensure that all sys-
tems were functioning properly. Two problems oc-
curred during this testing that would play a part
in the March 28, 1979, accident.
Problem with the Condensate Polishers
On October 19, 1977, a problem arose that was
almost identical to the one that triggered the 1979
accident. The outlet valves on the condensate pol-
ishers closed, and operators could not open the
bypass valve from the control room because the
control was inoperable. Instead, they had to open
it manually. (149) The manual control station for
the polisher bypass valve was, however, nearly in-
accessible, and it took great effort, in a physically
awkward position, to operate. (150) Later, it was
found that water had entered the air lines; this
was assumed to be the cause of the malfunctioning
of the outlet valves. (151)
Analysis of the Incident
John Brummer, a TMI-2 electrical engineer,
and Michael Ross, Unit 1 Supervisor of Operations
at the time of the March 1979 accident, analyzed
the event and prepared a memorandum. (152)
They discussed the problem of water getting into
the instrument air lines and suggested solutions to
preclude a recurrence. The memorandum also
stated a concern that if this malfunction were to
occur while the plant was at power, the emergency
feedwater system might be actuated, the turbine
would trip and the reactor might trip as well.
(153)
Brummer filed a problem report with R. J.
Toole, Manager for Startup Testing for GPU,
(154) to which he attached the memorandum. The
problem report focused on possible solutions to
the problem of water in the air lines and recom-
mended installation of an automatic bypass valve
in the system.
Response by Management
GPU did not implement this recommendation.
Ronald P. Warren, a member of the Plant Opera-
tions Review Committee, provided some insight
into that decision :
WARREN : I think it was because they said
it was a plant improvement that really didn't
have to be made.
Question : Didn't they say it costs too much ?
WARREN: They might have. That might
have been a better way of putting it. (155)
At the same time, GPU said that it would re-
evaluate the problem of water in the lines after
the plant had begun to produce power, in the be-
lief that the problem might have originated with
earlier flooding at the plant. (156)
Problems with the condensate polishers per-
sisted, 38 and at one point a full-time crew was as-
sembled with one responsibility to work on the
polishers. (157) Nevertheless, difficulties recurred.
In fact, a crew was working on the system when it
malfunctioned on March 28, 1979, initiating the
accident.
The utility did not inform the NRC of the prob-
lems with the condensate polishing system, and the
Office of Inspection and Enforcement did not
learn of the October 19, 1977 event until its in-
vestigation following the March 28, 1979 accident.
(158) However, the agency's reporting require-
ment applied only to defects believed to affect
safety. (159) According to the NRC, "problems
related to the condensate-feedwater system were
not considered by the licensee to be reportable be-
M By design, the polishers have to be changed periodically (every couple of days). One is taken out and recharged,
while another is rotated into its place. The procedure is difficult and has caused continuing problems, particularly in terms
of maintaining proper flow through the system.
64
cause the plant is designed to safely sustain a loss
of normal feedwater." (160) Burns and Roe in-
formed the Special Investigation by letter that
it had been responsible for the development of the
performance specifications and technical review
of the bids for the condensate polishing system.
The initial design provided that the outlet valves
in the system would fail "as is" if the air system
controlling them were to malfunction (e.g., if
water were to enter the air lines) : ". . . [the! spec-
ification [required! that the condensate polishing
system valves fail 'as-is' upon loss of either instru-
ment air or control [electrical! power." 39 (161)
Burns and Roe stated that the utilitv had not
told the company of the problems with the con-
densate polishing system on or after October 19,
lf>77. Burns and Roe did locate in its files a copy
of a GPU Problem Report concerning the October
19. 1977 occurrence, believed to have been sub-
mitted in late 1978 in connection with another
project. GPT had not asked Burns and Roe to take
any action, and Burns and Roe did not provide any
recommendations. (162)
Burns and Roe implied in its letter to the Spe-
cial Investigation that the valves may have failed
closed because of a design change of which it was
not informed. It wrote :
. . . all outlet valves of the condensate
polishing svstem had pll closed causing a
complete loss of feedwater flow. In one
case [this one!, the problem was asso-
ciated with water in the instrument air
lines . . . [This incident! indicate Fs] a
discrepancy between the actual perform-
ance of the condensate polishing system
and our specification requirement. . . ."
(163)
September 1977: Steam in the Hotlegs
At one point during hot functional testing 40 in
September 1977. steam became trapped in the hot-
legs. The primary system seemed to be filled with
water, but operators had difficulty establish-
ing natural circulation. Two operators on dif-
ferent shifts noted that the pressurizer level would
increase when the pressurizer was vented in order
to decrease primary system pressure. (164) This
l>ehavior was unexpected. (165) At least one oper-
ator suspected steam in the lines that measured the
pressurizer level. (166) An operator on yet an-
other shift deduced that the system not only had
steam in the measuring lines, but in the hotlegs as
well. (167)
The operator? eventually corrected the situation
by pumping nitrogen into the pressurizer. raising
the pressure sufficiently to force water from the
pressurizer into the hotlegs, thereby collapsing
the steam bubble. (168)
Babcock & Wilcox's site representative, 41 Le-
land Rogers, explained the problem to Special In-
vestigation staff :
... a phenomenon had occurred [in this
plant] where we had trapped a lot of
hot water [steam! in the hotlegs, and
subsequently had the rest of the system
colder. And without the ability to run
the reactor coolant pumps, which we did
not have at that time [during hot func-
tional testing! . we could not get the heat
out of those hotlegs; even with the sys-
tem filled with water, we could not move
any heat from that. It's in a natural
trapped condition. (169)
Rogers attributed the problem of stagnation
in the hotlegs to the "candy-cane" shape of the
lines:
It was accepted as a condition because of
the layout of the plant ... It has hap-
pened at other B&W plants, so it was not
a brand new problem. (170)
The behavior of the pressurizer and primary
system during this incident water level in the
pressurizer increasing as pressure decreased was
found to have occurred when steam formed in the
system. (171) When these same conditions ap-
peared on March 28. 1979, they were neither rec-
ognized nor understood because details of this
earlier incident apparently had not been com-
municated to operators on duty during the acci-
dent"
GOING CRITICAL
As noted, the TMI-2 reactor went critical on
March 28. 1978. It was the first day of the normal
operational testing phase that would continue for
several months. Over this time the plant gradually
would be brought up to full power and put in
service.
The unit experienced two minor accidents dur-
ing this period, both of which in retrospect were
significant in terms of the major accident that was
to come in March 1979.
MARCH 29, 197& A FUSE BLOWS
The dav after going critical, with the unit at
less than 1 percent power, a fuse blew in the plant's
electrical system, and power was lost to an elec-
" The post-accident investigations concluded that the valves did not fail "as is" on March 28, 1979.
* Hot functional testing, performed prior to initial loading of nuclear fuel, is designed to verify the ability of the
reactor coolant system to operate properly at pressures and tpmperatures comparable to those of normal operation.
" A vendor frequently will assign a full-time representative to a plant to assist in operations.
" See "The Accident at Three Mile Island : The First Day," pp. 97-98, 105-108.
trical control system associated with the pilot-
operated relief valve (PORV) on the pressurizer. 43
The system had been wired so that the PORV
would open automatically if power were lost. (172)
It did so, and water drained out of the primary
coolant system. Pressure dropped from 2,188 psi
to 1,173 psi. (173) The high pressure injection
system actuated automatically. (174)
Power was returned to the electrical system
about four minutes after the transient began, and
the PORV automatically closed. (175) Prior to
that point, however, the operators had not known
it was open because the TMI-2 control room did
not have an indicator showing the valve's position.
(176)
A PORV Indicator Installed
In response to this minor incident, the electrical
circuits were rewired so that the PORV would
remain closed in the event of a loss of power in the
electrical system. In addition, a position indicator
for the PORV was installed in the control room.
This indicator did not, however, provide a direct
indication of the position of the valve. Instead, it
was a command-type indicator that sensed whether
an electric command was being sent to the valve,
ordering it to open. If a signal to open were being
sent, then an indicator light in the control room
would be lit, suggesting to the operator that the
valve had opened. Conversely, the absence of a
light suggested it was closed. (177)
James Floyd discussed the request made after
the March 29 accident that a position indicator for
the PORV be installed. He stated :
Mot. Ed put in the problem report. We
asked for valve position indication, the
decision comes back from bosses that all
we're going to get is command valve com-
mand signal light. And he comes into the
control room and he has to sell me on that,
and if he can't sell me, then I'm going to
raise a stink and push for what I wanted
in the first place, or some compromise in
between his position and mine. (178)
Floyd later commented on his attitude toward
management's oversight and response to the
PORV issue. He said he "wasn't too happy" with
the decision to install an indirect indicator, but
added :
. . . my reaction was not to raise a big
stink and say, "Hey, I asked for valve
position indication. That's what I
want.". . . I was the type that is much
more inclined to do things on a low-key
basis and get them accomplished, rather
than making a big fuss and not getting
anywhere. (179)
Other TMI-2 operators interviewed indicated
that at the time they likewise accepted the decision
to install a command-type position indicator
rather than a direct indicator. According to Floyd
and Faust, an indirect indicator was better than
no indicator at all. (180) Zewe said that in a
"generic sense" a direct indicator' would have been
preferable, but that he never specifically raised the
issue of having one installed. (181)
Operators Requested a Change
Nine months after the March 29, 1978 accident,
in January 1979, the TMI-2 maintenance and en-
gineering staffs requested a design change in the
TMI-2 PORV position indicator. Joe Logan, the
Unit Superintendent, Richard Bensel, a lead en-
gineer, and Daniel Shovlin, Supervisor of Main-
tenance, signed the request. They proposed that
the indicator be modified so that a limit switch **
on the solenoid of the PORV would activate it.
Although this modification still would not have
provided a guaranteed indication of the PORV's
position, it would have provided a more reliable
indication than did the command-type indicator
then in use. (182)
The request for this change was forwarded for
review to the Met Ed Generation Engineering De-
partment in Reading. Pennsvlvania. Thnt depart-
ment disapproved the proposed modification. In a
March 16. 1979 memo to Shovlin. R, C. Noll of the
Generation Engineering staff said that "After dis-
cussing this modification with the TMI-2 staff, it
has been agreed that this modification is not neces-
sarv or required." (183) Bensel has stated that he
and George Kunder, the Unit 2 Superintendent
for Technical Support, had concluded that "the
added indication . . . wouldn't have been that much
better." (184)
Since it is still not known what part of the
PORV failed during the March 28, 1979 acci-
dent, it cannot be determined whether the pro-
posed modification would have shown that the
PORV was stuck open.
APRIL 1978: A MORE SEVERE PROBLEM
On April 23, 1978, the TMI-2 reactor expe-
rienced a more severe problem while at 30 per-
cent power. (185) According to a Met Ed analy-
sis, (186^ a minor equipment malfunction caused
the TMI-2 reactor and turbine to trip almost
simultaneously. When the turbine tripped, pres-
sure on the secondary side of the steam generators
41 See "How the Plant Works," p. 30, for a description of how the PORV operates.
44 The limit switch provides mechanical contact between the solenoid plunger and the solenoid housing when the
solenoid plunger moves into the valve-open position. The command-type indicator just shows that a signal has been sent to
move the solenoid plunger.
66
increased, and some of the main steam safety re-
lief valves opened. (187) Several failed to close,
and pressure in the secondary system dropped
rapidly. (188)
Two other factors complicated the problem.
First, the computer-controlled valves that regulate
flow in the feedwater lines closed much more
slowly than expected. (189) Second, the operators
failed to throttle the feedwater pumps until 1 min-
ute 20 seconds into this minor incident. (190) Thus
an excessive amount of water flowed to the steam
generators. (191) The result was rapid depressur-
ization and cooldown of the primary system. (192)
The coolant in that system contracted to such an
extent that the pressurizer was emptied of water.
A steam bubble then formed in the hotlegs of the
primary system. (193)
The operators started a second make-up pump,
and shortly thereafter the high pressure in-
jection system automatically activated. (194)
These responses restored the water level in the
pressurizer. and the operators eventually brought
the plant to a stable condition. (195)
Too Many Alarms
Effective operator response had been hampered
by a number of factors. Several hundred alarms
had activated so quicklv that the operators had to
ignore them temporarily and concentrate on the
gauges that measured kev plant conditions. (196)
Further, the computer that printed out the acti-
vation of alarms in sequence became backlogrged.
(197) In Frederick's words, the alarms "were not
being typed out fast enough to be of use in eval-
uating the transient." (198) Only after the oper-
ators had ascertained that the plant was stable did
they turn to the alarms.
Faust said that as a result of this minor accident.
he concluded that the alarm svstem would be "use-
ful" onlv if no more than three or four alarms
were activated. (199) If more came on. ". . . then
you had to nick through them to find the ones that
were significant. It took too much time." (200)
That had been his experience on April 23 "it
was just useless to try to really sort out all the
alarms that were coming in.'' (201)
Frederick agreed : "I was forced to ignore most
of the alarms." and commented that he had to rely
on the information presented by the gauges on the
control panels. (202) He noted:
. . . with that much alarm information.
the information begins to lose its value . . .
the flood of information has no priority
and no time sequence. (203)
Analysis of the Accident
Shortly after the April 23, 1978 accident, James
Seelinger, the Technical Support Superintendent
for TMI-2. prepared a detailed analysis of it.
(204) He identified a number of equipment fail-
ures that had contributed to the incident, includ-
ing failures or deficiencies in the main steam
safety relief valves, the main feedwater block
valves and the emergency feedwater control valves.
(205)
In addition, although he praised the operators
for their response to the reactor trip and actuation
of high pressure injection, he was critical of two
other aspects of their performance: their failure
to slow the feedwater pumps sooner and to diag-
nose the accident correctly. (206) As a result of the
latter failure, he concluded that some of their re-
sponses had been inappropriate :
While the operators responded correctly
to the reactor trip, they did not realize the
casualty they were really dealing with
was a major steam leak (through the re-
lief valves).
The operators during the transient
never fully grasped the damaging effect
of feedwater on this situation. . . . (207)
In his report. Seelinger noted that indicators of
several "critical operator items" were needed in
the control room and recommended they be in-
stalled. (208) They included position indicators
for the main steam safety relief valves, feedwater
block valves and emergency feedwater control
valves, (209) However, in his report Seelinger
never associated the lack of indicators on the po-
sitions of the valves with the operators' inappro-
priate responses. (210)
Seelinger recommended that an effort be made
to reduce the number of "nuisance"' alarms that
would activate while the plant was operating nor-
mally. 45 (217) He also recommended that addi-
tional alarm acknowledgment stations be added in
the control room. (218) His report did not dis-
cuss whether these features had contributed to the
incident.
Response by the Operators
Seelinger's analysis was distributed to control
room operators at TM 1-2. Three of those present
in the control room during the accident Ed
Frederick. Hugh McGovern and Craig Faust
discussed the report and the accident. (219)
Frederick responded to Seelinger's criticisms,
(220) Although Frederick said his letter to
* Nuisance alarms are alarms that remain activated in the control rnom during periods of normal operations. (211)
They can he actirated as a result of faulty wiring or if the sensors that trigger the alarms are overly sensitive. (212) In
addition, some alarms are wired to activate in response to routine actions of operators, such as turning off a pump. (213)
Many can remain lit for extended periods. (214)
Zewe stated that in 1978 at times over 100 "nuisance alarms" would be lit in the TMI-2 control room. Electrical
engineer John Brummer estimated the average was about 70. (215) For comparative purposes, the TMI-1 control room
typically had about five nuisance alarms lit. (216)
67
Seelinger represented only his views, Faust said
that Frederick had showed it to him and that he
agreed with it. (221) In Faust's opinion, See-
linger's analysis
. . . seemed to be pointing a heavy finger
at the operators as the cause of some of
the problems we had at the time in that
transient and we were just saying that
not all the indications that we could have
used were available to us at the time. . . .
(222)
Frederick said he wrote Seelinger on May 3,
1978 in order to identify "the problems that I saw
in the accident, that I didn't think were touched
by his evaluation." (223) In his letter he com-
mented on the alarm system :
The alarm svstem in the control room is
so poorly designed that it contributed
little in analysis of a casualty. The other
operators and myself have several sug-
gestions on how to improve our alarm
system perhaps we can discuss them
sometime preferably before the system
as it is causes severe problems. (224)
Frederick said he intended that Seelinger would
assign an engineer to work with the operators "on
a long-term basis" to correct the problems. (225)
Elsewhere in his letter Frederick challenged
Seelinger's analysis of the causes of the accident :
I feel that the mechanical failure, poor
system designs, and improperly prepared
control systems were very much more the
mai'or cause of this incident than was op-
erator action.
You might do well to remember that
this is only the tip of the iceberg. Inci-
dents like this are easy to get into and
the best operators in the world can't
compensate for multiple casualties which
are complicated by mechanical and con-
trol failures. (226)
Frederick said that by "improperly prepared
control systems," he meant the absence of several
important valve position indicators in the con-
trol room. (227) In Frederick's words. "We
couldn't see all the valves that were necessary to
control the system." (228)
Frederick closed his letter with the following
comments :
Some of our suggestions are good. We
made siurcrestions on FW [feedwaterl
valve indication 2 years ago (submitted
many FCR's) . We have complained about
the alarm system since day one. Let's get
together and try to prevent this from
happening again. 46 (229)
Seelinger replied to Frederick's letter the same
day. (230) He reemphasized that the operators'
response had been "good and proper." (231) As
to the alarm svstem, he stated that Frederick's con-
cerns were addressed by the recommendations in
Seelinger's report on the accident. (232) In re-
sponse to Frederick's "tip of the iceberg" com-
ment and his concerns about the difficulties op-
erators had in responding to accidents such as that
of April 23. 1978, Seelincrer commented, "the abil-
ity to do this comes with experience, and I think
the operators who had this transient performed
well considering their experience." (233)
Frederick told Special Investigation staff that
he believed he was "fairly pacified" by Seelinger's
response. (234) However. Fanst said that he and
Frederick "didn't exactly like the response we got
back." (235) Faust added that at the time. Fred-
erick thought that "what we were asking about or
mentioning just wasn't going to have too much fol-
lowup. . . ." (236)
According to Frederick, after receiving See-
linger's response, he had one brief, informal dis-
cussion with him about their letters and See-
linger's analysis. (237) However, they "never had
the meeting that I requested about the alarm sys-
tem." (238)
Design Modifications
TMI-2 was shut down for approximately four
months after the April 23 accident. (239) During
that period, the utility tested and ultimately re-
placed the main steam safety relief valves. (240)
TMI-2 control room operators said that their
emergency procedures were also revised in an at-
tempt to ensure they would not overfeed the steam
generators in the event of a similar accident. (241)
The revised procedures and the April 23 incident
were discussed in operator training programs at
TMI. (242)
With respect to control room instrumentation,
the utility installed position indicators in the con-
trol room for the valves in the feedwater system.
(243) In addition, it set up a microphone in the
vicinitv of the main steam safety relief valves and
wired it to a sneaker in the control room so that
onerators could hear when the valves were open.
(244)
Changes in the Alarm System
As to the alarm system, Floyd said that after the
accident, the operations staff requested the instal-
lation of several additional alarm acknowledge-
ment buttons in the control room. (245) They
wanted each to acknowledge only the alarms in a
certain section of the control room. (246) This
would lessen the chance of an operator inadvert-
ently acknowledging new alarms coming into the
control room during a transient while clearing
'FCR's Field Change Requests nre one of the formal means by which operators can request design changes.
68
previously acknowledged alarms, since he would be
working only with one section of the control room
at a time.
The utility installed several additional buttons,
but they still acknowledged all the alarms in the
control room. (247) Management's rationale was
that this system would be easier for operators
moving around the control room. (248)
Evidence reviewed by the Special Investigation
indicates that the addition of the acknowledge-
ment buttons was the only design change in the
TMI-2 alarm system made specifically in response
to the April 23 accident. (249) Beyond that, in late
1978 and early 1979. Met Ed and B&W engineers
did reduce the number of "nuisance alarms." as
Seelinger had recommended. Nevertheless, just be-
fore the March 28. 1979 accident began, about 50
of those alarms were activated. (250)
Operator* Dissatisfied
Even with these modifications, operators were
still dissatisfied with the alarm system. Floyd said
that no effort had been made to restructure the
alarm system so that the operators could identify
and react quickly to the most important alarms.
(251) He concluded that such an approach
"wasn't recognized as being necessary" and added.
"I don't know that it was really a recognized need
until this transient on March 28, [1979].* (252)
Alarms To Be Ignored
One result of their dissatisfaction with the alarm
system was that Faust. Frederick and Zewe said
they would not acknowledge alarms activated dur-
ing the initial stages of a complicated transient.
(-253) They knew that they would not be able to
read all the alarms and respond to the situation
at the same time. If they wore to acknowledge the
alarms without first reading them, they might in-
advertently clear some without noticing they had
been activated. (254)
Tn an interview with Special Investigation staff.
Floyd confirmed that the TMI-2 operators had de-
cided not to acknowledge anv alarms during the
early stages of a transient. He said he considered
this approach to lie acceptable, since all the occur-
rences prior to March 28. 1979 had been of very
short duration :
. . . the reactor trips and it's all over.
Thirty seconds, and it takes the operator
two minutes to realize that it's over. And
then he can scan his board and put the
plant back together again. (255)
Floyd did not specifically recall telling oper-
ators after the April 23 accident to "ignore" the
alarms during a mai'or transient, although he
said. "I certainly wouldn't be surprised if I did."
(256) He commented :
If the man came up to me and said, "Hey,
the alarms are absolutely worthless dur-
ing this transient." I would say, "Yes,
they always are, but you got these meters
and recorders over here. They are what
are going to keep you out of trouble. Go
ahead and ignore those overhead alarms,
but pay attention down here. This is
where it's at." (257)
In Zewe's opinion, the alarms would not, in any
event, play a "major role" in operator efforts to
diagnose a major accident :
. . . I've been involved in several tran-
sients and we handled them pretty much
the same, whereas we didn't use the
alarms and accepted what we had and we
just used it to the best of our ability.
(258)
Zewe did. however, note one problem with fail-
ing to acknowledge the alarms as they were acti-
vated. (259) The operators would not know the
sequence of their activation. (260) Given that
the computer alarm printer would become back-
logged during a major transient, 47 there would be
no way to obtain the complete alarm sequence,
probably until the transient was over.
The ability to reconstruct a sequence of alarms
is an important analytical tool for an operator
faced with a complex transient. As Frederick
noted.
... if [an operator] makes a misjudg-
ment early on in the accident, he has the
ability to review the effect of that mis-
judgment on the trend as a whole and
retrace the steps: start a more logical
strain, have more logical information
[on which] to base his decisions. (261)
On March 29. 1979 the operators would have the
same problems with the alarms as they had dur-
ing the March 1978 accident. Frederick noted that
most operators simply did not use the visiial and
computer alarm systems in the early stages of a
transient :
The general consensus and the training
that we received, that led most of the op-
erators to not use the computer as a source
of information during a transient, you
could use it after things settled down, but
during the first phase of any transient,
you would just have to go by the infor-
mation available on the panel, since
neither the [annunciator] nor the com-
puter will be able to give vou good time
sequence and prioritized alarm informa-
tion. (262)
' Computer alarm deficiencies are discussed in greater detail on p. 71.
69
According to Frederick, he was also concerned
about the alarms on the back control panels that
were completely out of sight of the main con-
trol console. He noted that if the alarms on the
front and the back control panels were to come on
simultaneously, the operators could acknowledge
the front panel alarms without necessarily notic-
ing those on the back panel. (263)
Frederick stated that he believed the TMI-2
control room alarm system could have been de-
signed to be useful during a major accident:
We need an alarm system that is designed
to be useful during analyzing one of these
problems. We need alarms that are mean-
ingful. In other words, you need an alarm
that tells you when you have lost a feed
pump, but you don't need one that tells
you that there is trouble in the turbine
building elevator. Out of the 1200 or 1600
alarms that are displayed up there, I am
sure we could narrow that down to 100
or 200 without losing any vital indica-
tions. The need to acknowledge wouldn't
be necessary. (264)
Thus, prior to the accident on March 28, it is
clear that the operators had decided that the alarm
system would not be of any immediate assistance
in a major transient or an accident. On March 28
the operators acknowledged the alarms in the early
moments to silence them, but did not read them. 48
The Usefulness of an Alarm System
In their statements, several operators suggested
that an alarm system was not really important in
an accident. Floyd made a number of comments
on this issue. He said that the March 28, 1979 ac-
cident made him realize that operators cannot
themselves handle the "information overload"
created by the activation of several hundred alarms
during the first minutes of a transient. The op-
erator :
. . . needs the prioritization to take place
automatically for him . . . you have to
give the operator some assistance to sort
out those several hundred alarms which
are coming in so very rapidly, all of them
calling for his attention, when, in fact, he
should only be paying attention to a half
dozen of them. (265)
At the same time, Floyd recognized that the
operator would have to work with his other instru-
mentation and equipment :
There's an information overload during
a transient that the control room operator
has to live through. And neither the an-
nunciators nor . . . the high-speed printer
is going to get him out of that difficulty.
The information overload is going to be
there and he lias to go back and rely on
his console indication on the meters, on
temperature, pressure, flow and power . . .
Whether it's a short-term transient or a
long-term transient, he's got to pay atten-
tion to those big four. That's funda-
mental. (266)
Floyd elaborated on this point. Even if the
alarm system were structured so that operators
could react quickly to important alarms,
. . . the operator is still not going to run
his control room based on the alarms in
the transient ; he's going to run the control
room based on the panel indication. He
has to. That's the only thing that's in the
real time. (267)
He added that
. . . you don't want to try to cure all the
operator's problems with a sophisticated
alarm system . . . [although] you can
do some additional sophistication that we
don't have, to get him pointed in the right
direction, to maybe help explain what
those four big meters are supposed to be
telling him already in the console. (268)
When Zewe was asked if the instrumentation
had caused the failure of control room personnel
to diagnose the stuck open PORV on March 28,
1979, he replied :
It is pretty hard to say. but I think it is
a matter of a combination [of things].
I think you could pick any certain area
and say, "Yes, we should improve that
area and instrumentation, improve the
procedures we use. improve this or look
at something else." All of that certainly
had a factor and to what degree depends
on whose evaluation it is. So I could ra-
tionalize and say, "Yes, I think there were
many things we could have done better,"
certainly without a doubt. And through
every instrument we had [.] I think wo
had all the mechanisms we needed to cope
with the situation at this point, but there
were certainly things that could be im-
proved upon that would cause more
awareness or recognition of the problem;
yes. (269)
He concluded :
I feel the plant is of a good design, that it
is adequate. I feel the training was good
and adequate. There again, I felt the in-
strumentation Avas adequate, too. Like I
' See "The Accident at Three Mile Island : The First Day," p. 99.
70
mentioned before, everything needs fur-
ther improvement. We can always use
that (270)
OTHER PROBLEMS AT TMI-2
Control Room Computers
Some alarms are not represented by flashing
lights in the control room, but instead appear only
on the computer printout. (271) One such alarm
that played a role in the March 28, 1979 accident
was the alarm indicating a high water level in the
containment sump. (272)
The computer in the TMI-2 control room is a
Bailey 855. (273) It has two printers similar in
appearance to electric typewriters. (274) One
prints out system conditions on request : the other
prints out alarm information, at a rate of approxi-
mately 14 alarm messages per minute. (275) If the
alarms are activated faster than that, they are
stored in sequence in a memory bank that can hold
a total of 1.365 alarm messages. (276) The com-
puter will automatically print out the alarm mes-
sages in the memory bank until the backlog is re-
duced to zero. ( 277 1
Scheimann. Faust and Frederick noted prior
to the March 28 accident that the computer alarm
printer would become backlogged during routine
reactor and turbine trips. (278) Frederick stated
that it "can be as much as an hour behind on just
the turbine trip." (279) Zewe likewise said he
knew it would become backlogged during "major"
transients. (280) Faust and Frederick both stated
that the printer became backlogged during the
April 23. 1978 transient. (281)
As noted, determining the sequence of alarms
can be important to diagnosing an accident and re-
sponding properly. The computer could be of as-
sistance in that determination. Because of the back-
log problem, however, operators recognized that
the TMI-2 computer would be of little help. In
Zewe's words.
During any major transient. I would to-
tally ignore the alarm typewriter, know-
ing from past experience that it would
backlog and I wasn't interested or con-
cerned in past history, more so in current
parameters. So I would only use the com-
puter for current parameters or until it
caught up to the real time frame of the
alarm, then I would start to use it in a
normal one-on-one fashion. (282)
Leakage of the Pilot-Operated Relief Valve
Since October 1978. the pilot-operated relief
valve (PORV) had been leaking. (283) as indi-
cated by high temperatures in the discharge line
leading from the valve to the reactor coolant drain
tank. Temperatures, usually a little below 130 F.
were averaging around 170 F or 180 F during
normal operation, a little below the 200 F point at
which an alarm indicating a high temperature
sounds in the control room. Prior to the 1979 acci-
dent, plant personnel were aware of some leakage,
but were not certain whether it was the PORV or
one of the two code safety valves on the pressur-
izer. (284)
Emergency Procedure 2202-1.5. "Pressurizer
System Failure," requires that plant personnel re-
spond to suspected PORV leakage by closing a
"block" valve located in front of the PORV and to
suspected code safety valve leaka'ge by recording
their discharge line temperatures on an analog
trend recorder, if temperatures exceed 130 F.
(285)
The utility had neither closed the block valve,
nor used the recorder, nor repaired the leak-
ing valve prior to the March 29. 1979 accident.
(286) Frederick said there had been some con-
sideration of closing the block valve: (287) but
this was not done because of the possibility of its
sticking in the closed position. 49 (288)
The actions of the control room personnel on
March 28. 1979 indicate that they had become ac-
customed to the elevated temperatures produced
by the leakage. That, combined with their knowl-
edge that the PORV had lifted briefly, releasing
more hot steam and water and raising the tem-
perature still further, and the fact that the indirect
indicator light in the control room went out. ac-
cording to the control room personnel, misled
them into thinking the valve was closed. (289) For
more than two hours, they did not recognize that
the PORV was stuck open. 50 (290)
Inspectors from the XRC's regional office had
conducted 25 inspections of varying types at
TMI-2 between October 1978 and the March 1979
accident. The inspectors never noted the leakage
from the PORV. (291) The XRC explained that
the alarm setpoint of 200 F was never reached,
and that an inspector doing an audit type of in-
spection would not have been expected to note the
elevated temperature, unless the alarm had gone
off. (292)
See "Recovery at Three Mile Island." pp. 210-211. Most of the civil penalty of $155.000 the XRC assessed Met Ed for
violations can be attributed to the utility's failure to respond to the leakage as required.
In connection with this same leakage, in 1980 the XRC asked the Department of Justice to investigate allegations by
Harold Hartman. a former TMI control room operator, that the utility had manipulated the calculations of the rate of
leakage in order to get figures within the limit set by the Technical Specification governing plant operations. According to
Hartman. plant personnel added water to the coolant in the primary system and altered the hydrogen overpressure in the
make-up tank but did not include those changes in conditions in the calculations.
" See "The Accident at Three Mile Island : The First Day." pp. 95-109.
71
If the leakage had been severe enough to neces-
sitate repairing the PORV, the utility would have
to have shut the plant down. In that event, the
utility would have been required to notify the NRC
of the leakage. (293)
Frequent Actuation of High Pressure Injection
Actuation of the high pressure injection (HPI)
system indicates a potentially severe problem,
since it is designed to replace coolant lost during
a small loss-of-coolant accident. The system auto-
matically comes on when primary system pressure
falls to 1,640 psi, ordinarily reached only during
a loss-of-coolant accident. (294)
At TMI-2, however, the HPI system had actu-
ated four times in the 12 months prior to March
1979 in response to relatively routine problems in
the secondary system, rather than to loss-of-cool-
ant conditions. 51 (295) TMI-2 operators had be-
come accustomed to initiation of HPI under less
severe conditions. In the words of TMI shift super-
visor Ken Bryan, "It is not a big deal" for the
HPI system to activate at TMI-2 during a tur-
bine/reactor trip. (296)
The NRC Office of Inspection and Enforcement
noted in. its investigative report of the accident
that "the operators were not surprised by HPI
actuation." (297) Further, "the operators were
conditioned to promptly bypass ES [engineered
safety! without first determining the condition of
the RCS [reactor coolant svsteml." 52 (299)
It can be inferred from the NRC's findings that,
on the day of the accident, the control room person-
nel throttled the HPI severely without determin-
ing if there were a loss of coolant because of their
past experience with its initiation.
The utility had reported the previous HPI ac-
tivations to the NRC. (300) Each was, in turn,
reported in NTJREG-0020, "Operating Unit
Status Report," better known as the Grey Book,
and was analyzed by personnel in the regional of-
fices. (301) The analysis of the HPI activations
focused, however, on the effect of the addition of
sodium hydroxide, a chemical added to the in-
jected water, on the system, rather than on the
fact that HPI came on for other than loss-of-
coolant conditions. (302)
In response to the HPI problem, the low pres-
sure reactor trip setpoint was raised 100 psi to
1,900 psi and the point at which HPI could be
bypassed was raised by 100 psi to help reduce un-
necessary actuations. (303) The NRC issued no
special notification to other utilities of the prob-
lem. (304)
Babcock & Wilcox also looked into the actuation
of HPI at TMI, but took no action to limit its ac-
tivation under non-loss-of-coolant conditions,
(305) nor did it notify other utilities with B&W
systems of the problem. (306)
The Special Investigation staff discussed with
the NRC the possibility that operators might se-
cure HPI in a loss-of-coolant accident if they were
accustomed to the system starting in response to
feedwater problems. (307) An NRC official ex-
pressed concern, although for a different reason.
(308) In a loss-of-coolant accident, the reactor
coolant pumps should be shut off to slow the rate
at which coolant is lost from the system. NRC staff
is seeking unambiguous signs that will alert the
operator to a loss of coolant so that the operator
will know to shut the pumps off. One sign could be
HPI, but only if the system were to come on just
for losses of coolant. (309) NRC staff has stated
that the Special Investigation staff's concern about
operator conditioning is valid and complementary
to its concern. (310)
IN RETROSPECT
During the three years from 1976 until the
March 1979 accident, TMI operators and super-
visors became aware of several weaknesses in the
design of the control room and its instrumenta-
tion. Most of those problems became apparent dur-
ing several minor accidents that occurred at the
plant during operational testing in 1978. Many
played a role in the March 1979 accident. However,
according to Zewe, although instrumentation and
procedures could have been improved, the person-
nel "had all the mechanisms we needed to cope
with the situation" on March 28, 1979. (311)
The history of TMI-2 provides a good exam-
ple of the consequences of the lack of attention
given to human factors engineering in control
room design. In various events that occurred dur-
ing early operations, the TMI-2 control room
alarm system overwhelmed operators with infor-
mation and failed to assist them in diagnosing the
situation. Although the deficiencies of the alarm
system were of obvious concern to several control
room operators, the TMI-2 Supervisor of Opera-
tions was less concerned, since he believed that
during a transient the operators would have to
concentrate on indicators of plant conditions, and
not on the alarms. Others shared his opinion.
Control room personnel had identified other
problems as well. These problems also influenced
events on the day of the accident. They involved :
1. The condensate polishing system;
2. The configuration of the hotleg piping,
which could trap steam, blocking the flow of
coolant and causing unusual plant behavior;
81 The March 29, 1978 incident evolved into a loss of coolant accident after the POKV stuck open.
52 The report also said that "the normol course of n>ost ES initiations, those which did not involve a loss of coolant,
required bypassing of ES and securing of HPI. . . ." (298)
72
3. The leakage through the pilot -operated
relief valve : and
4. The actuation of the high pressure in-
jection system under non-loss-of -coolant con-
ditions.
Neither the utility, its suppliers nor the XKC re-
sponded to these problems in a way that effectively
prevented their recurrence.
OPERATOR TRAINING AND LICENSING
A critical element in the safe operation of nu-
clear powerplants is the preparedness of plant
personnel, particularly the operators and supervi-
sors. As became evident on March 28, 1979, the
TMI-2 operators and supervisors were not ade-
quately prepared to diagnose and respond to the
accident. In light of this, the Special Investigation
reviewed briefly the training provided to utility
personnel. 53
In general, operator training at TMT empha-
sized plant operations under normal conditions and
response onlv to selected "standard" accidents. Op-
erator? had limited instruction or practice in diag-
nosing and responding to multiple failure acci-
dents, particularly prolonged ones, such as oc-
curred on March 28.
Further, their emergencv procedures, which
thev had been trained to use in unusual situations,
did not provide needed guidance in the first hours
of the accident. In addition, the operators had in-
sufficient instruction on the basics of nuclear
powerplant phvsics and behavior. This contrib-
uted to the difficulty they had in diagnosing and
resnonding to the accident.
Although the utility is largely responsible for
the inadeo^iacy of operator training, the reactor
vendor, which helped develop the training pro-
gram, and the XRC. whose involvement in train-
ing was too limited, are also responsible for the
inadequacy.
OPERATOR TRAINING
Operators are required to have specialized edu-
cation and training and to be licensed by the NTJC.
Training i= conducted by the utility, frequently in
com'unction with the suppliers of plant systems.
In the case of Unit 2 operators. Met Ed had con-
tracted with Babcock & Wilcox to provide certain
portion? of the operator training program. (312)
The contract called for classroom and simulator
instruction for trainees. The courses were both
developed and taught by the B&TV training de-
partment. (313)
Weaknesses in the Training Program
There were several significant weaknesses in the
TMI operator training program that made it dif-
ficult on March 28, 1979, for the control room per-
sonnel to understand and respond to the sequence
of events.
Orientation of Training
For the most part, training was geared to nor-
mal plant operations and to the hypothetical acci-
dents postulated in the Final Safety Analysis Ke-
port submitted to the AEC. (314) Of significance
in the context of the TMI accident, the program
included only limited training in multiple-fail-
ure events and events of prolonged duration. (315)
Emergency Procedures
In addition, although the operators had been
trained extensively with the emergency proce-
dures, none of those procedures precisely antici-
pated the actual chain of events at TMI-2. 54
Faust stated :
That is what we were having a big prob-
lem with that day, trying to follow limits
set forth in our tech specs [Technical
Specifications], as well as our Emergency
Procedures, where we were having a diffi-
cult time doing that because we saw some-
thing diverse, something different from
what our training had taught us in the
past. (316)
Further, the sequence of events at the start of
the accident was much different than what the op-
erators had studied and from what they expected.
Zewe commented on his training for loss of cool-
ant accidents:
... If you look at any LOCA Floss-of-
coolant accident] we've ever had, if you
have . . . pressure in the building, and
also reactor coolant system pressure,
they're within seconds of each other.
(317)
During the accident these events seemed dis-
jointed to the operators. 55
Duration of Accidents
The TMI-2 accident also lasted longer than
anticipated. The operators found it difficult to
reconstruct events over that extended period and
to assess their evolution. Instead, operators fo-
cused on the state of the reactor at a given moment.
As Scheimann stated :
In the event that we had an emergency
that didn't fall within the scope of an
emergency procedure, the thing we would
The President's Commission and the XRC Special Inquiry examined operator training extensively. The Special
Investigation independently reviewed their materials and findings, as well as materials and interviews it compiled.
14 It should be noted that written procedures are based on certain foreseeable circumstances and are not meant to
cover all possible situations or to substitute for operator training in responding to unforeseen situations.
* See "The Accident at Three Mile Island : The First Day." pp. 102-103.
73
54-058 0-80-6
do would be to treat the symptoms, in
other words, respond to what we were
seeing in front of us. If pressure were de-
creasing, we would try to increase pres-
sure, and vice versa. . . . (318)
Basics of Plant Operations
Training was deficient in the basics of nuclear
powerplant physics and behavior. As is detailed
in the next chapter, the unusual behavior of the
plant wa's neither understood nor quickly diag-
nosed. For example, control room personnel failed
to appreciate the significance of the data on pres-
sure and temperature indicating saturated steam
conditions in the core. Marshall Beers, Met Ed
Group Supervisor of Nuclear Training, explained
that saturation conditions in the core were not
anticipated as long as pressurizer level was main-
tained. He added that, prior to the March 28 acci-
dent, operators were not trained for saturation
conditions in the core, although in hindsight these
conditions might possibly have existed during
previous reactor trips at TMI. (319) He further
noted :
. . . the significance of temperature-
pressure and the possibility of uncover-
ing the core under these types of condi-
tions . . . was never specifically taught
[to the operators]. (320)
In addition, while several of the control room
personnel recognized as early as 10 a.m. on the
first day of the accident that superheated condi-
tions were present, the Special Investigation staff
found no evidence that they linked that condition
to core uncovering.
Pressurizer Level
Operators had not been instructed that under
certain conditions the pressurizer level would be
an unreliable indicator of water level in the re-
actor vessel. In addition, their training led them
to believe that as long as there was adequate water
in the pressurizer, there had to be adequate cool-
ing water around the core. (321) This belief , along
with the standing instruction never to permit the
pressurizer to "go solid" (fill completely with
water) during plant operations, led the operators
to throttle high pressure injection early in the ac-
cident. (322)'
Zewe, a TMI-2 operator who was on duty the
day of the accident, said he was confused over why
the pressurizer level stayed high in the earlv hours
of the accident, even though coolant level should
have been decreasing as a result of the throttling
of high pressure injection and increased drainage
of coolant through the let-clown system. (323) Sim-
ilarly, James Floyd, TMI-2 Operation Supervi-
sor, said, ". . . to see the pressurizer level high and
the plant pressure low was just a situation that . . .
was never prepared for." (324)
Use of Instrumentation
The operators had not been familiarized with
the use of the movable incore detector, a com-
ponent of the plant's nuclear instrumentation that
could have been used as an alternative means of
detecting core uncovery. (325) Special Investiga-
tion staff interviews revealed that some utility em-
ployees who might have been expected to be knowl-
edgeable about the use of the detector Met Ed's
instrumentation foremen and technicians did not
even consider the device to be the property of the
utility :
... As far as I was concerned, it was not
part of our equipment. The only time I
have seen it in operation was when B&W
people were moving it up and down. The
Metropolitan Edison instrument men
never had anything to do with it. It was in
a separate cabinet all to itself over by the
side. (326)
Other utility personnel had never seen the de-
tector used, and no one the Special Investigation
staff interviewed recalled its being discussed or
considered on March 28, 1979. 56 (327)
Further, the control room personnel were not
trained to use the fixed neutron detectors to de-
termine water level in the core. 57 (328)
The remaining nuclear instrumentation avail-
able on March 28, 1979 were the source and inter-
mediate range monitors, which measure neutron
activity in the core. The operators were not
trained to use them to determine if the core is un-
covered. (329) Rather, they were instructed to use
the monitors when bringing the reactor back to
full power following a shutdown (for this rea-
son they are referred to as "start-up" monitors). 58
(330)
The Simulator
An important part of the training program in-
volved practice on a reactor simulator. This was
done at the B&W facility in Lynchburg, Va.
The simulator is a computerized mock con-
trol room that can reproduce events that occur in
a nuclear powerplant. However, the B&W simula-
tor differed in significant ways from the control
room at TMI-2. (331) For example, it was smaller
M See "The Accident at Three Mile Island : The First Day," p. 112.
" Part of the reason for not relying on the fixed incore neutron detectors is that under normal operating conditions,
accurate readings are produced by the control room computer only if the reactor is above 15 percent power. When a
reactor is shut down following a reactor "trip," normally there would not be enough current for the computer to produce
readings of neutron activity in the core. As a result, the operators tended not to rely on the neutron detectors when the
reactor was at a low power.
* See "The Accident at Three Mile Island : The First Day," pp. 111-112, 117-118, for further details.
74
The Babcock & Wilcox simulator, used in training operators of Three Mile Island
and more compact. It did not have nearly as many
alarms (150 vs. 1,200), and all the electrical distri-
bution instrumentation was on one panel, instead
of taking up one-third of all the panel space, as
in the control room at TMI-2. (332) Further, the
simulator was not programmed to reproduce all
the emergency conditions an operator might pos-
sibly have to address, (333) including the sequence
of eVents experienced during the March 28, 1979
accident.
As was noted, the program did not include ex-
tensive training in multiple failure events. Craig
Faust and William Zewe, two TMI-2 operators
who were on duty March 28, 1979, commented on
their simulator training in this area. Faust
said, ". . . we didn't train for multiple casual-
ties." (334) Zewe said that prior to the accident,
he had been trained "only to a limited extent" in
multiple failures at the B&W simulator. (335) He
added. "Maybe we would have two failures or one
failure and we caused another one by how we re-
acted to it. but not to the extent which they train
now, to where they will give you . . . several fail-
ures in a row, not like that, no." (336)
The lack of attention to multiple failure events
was the result in part of the NRC's single failure
criterion. According to this criterion, the licensee
only had to assume a limited number of concurrent
failures in the analysis of certain accidents. 59 Its
operators would be trained accordingly. (337)
OPERATOR LICENSING
On completion of their training, operators must
be licensed by the NRC before they can operate the
plant. They must pass oral and written examina-
tions administered by the Operator Licensing
Branch (OLE) of the NRC's Office of Nuclear
Reactor Regulation, the unit responsible for the
operator licensing program. (338) The TMI-2
operators on duty on March 28 had all scored above
average on the exams.
The NRC's licensing requirements for operator
training are contained in Part 55 of Title 10 of the
Code of Federal Regulations. 60 These provisions
59 These events are spelled out in Chapter 15 of the Safety Analysis Report.
* In addition to the regulations, NRC has two Regulatory Guides that address operator training. Regulatory Guide
1.8, "Personnel Selection and Training," endorses ANSI Standard 18.1, "Selection and Training of Nuclear Power Plant
Personnel." This standard outlines criteria for the selection, training, qualifications and responsibilities of operating
personnel. The standard was redrafted and circulated for comment by the American Nuclear Society as ANS 3.1 just be-
fore the accident at TMI. Following the accident, the draft standard was recalled and revised. The newly drafted version
is to be issued for comment in July 1980.
Regulatory Guide 1.70, "Standard Format and Content of Safety Analysis Reports for Nuclear Power Plants,"
provides guidance regarding information that is to be submitted to the NRC on training programs for plant staff. The
plans are reviewed by the Operator Licensing Branch, using the criteria contained in NUREG-75/087, Standard Review
Plan, Section 13.2, "Training."
75
establish two types of licenses: an operator's li-
cense for personnel who handle the reactor con-
trols ; and a senior operator's license for those who
supervise or direct the activities of the control
room.
As outlined in the American National Stand-
ards Institute Standard 18.1, an operator must
have a high school diploma or its equivalent and
have two vears of experience at a powerplant or
its equivalent, with a minimum of one year at a
nuclear powerplant. A senior reactor operator
must have a high school diploma or its equivalent
and four years of nowerplant experience in a posi-
tion of responsibility. A maximum of two years
of experience can be fulfilled by academic or re-
lated technical training.
There are no NRC requirements for psvchologi-
cal evaluation of license applicants. (339) nor is
there any investigation to determine the appli-
cant's employment history or whether he has a
criminal record. (340)
The NEC's role in operator training has been
quite limited and has principally involved audit-
ing the training programs. (341) It has set no
qualifications for the instructors who carry out
operator training, (342) and has no requirements
that training include proper response to signifi-
cant transients that have occurred at nuclear re-
actors. (343)
Requalification
In addition to the initial operator training, the
NRC requires that a utility conduct annual re-
qualification programs for its licensed operators.
Reaualification is actually continued training, in-
tended to ensure that licensed operators maintain
their technical skills and are aware of new pro-
cedures. Biannually the NRC audits the program
by checking the contents of examinations. (344)
The NRC's Role
Paul F. Collins, Chief of the Operator Licens-
ing Branch of the NRC, described some of the
shortcomings of the NRC program:
-In the written requalifvinqr examination,
only two of the seven or eight parts con-
tain questions on procedures relating to
safety and emergency equipment. An op-
erator could do poorly in these areas and
still achieve an 80 percent score overall,
thereby qualifying for license renewal.
(345)
-The NRC does not audit the requalifica-
tion training program administered by
vendors and -utilities for operators placed
in accelerated training after scoring less
than 70 percent on the written examina-
tions, if they perform well on the oral ex-
aminations. (346)
-Since 1973, the NRC has not required an
applicant for a new license to start up the
reactor in the presence of an NRC exam-
iner. Perhaps once a year. NRC examiners
observe students perform this task on a sim-
ulator. However, the NRC does not audit
the requalification training on simulators.
(347)
-If two units are sufficiently similar, an op-
erator licensed on one unit may be cross-
licensed for the other upon completion of a
"differences" course and an examination ad-
ministered bv the utility. A utility that
wants to cross-license its operators is
entirely responsible for that program. An
NRC examination is not required, and the
NRC dops not audit those given by the
utility. (348)
-The Operator Licensing Branch does not
coordinate its work with that of the NRC
staff who review desijrn aspects of a plant.
Thus, there is no direct communication
within the NRC on issues involving "where
man and machine come together." (349)
-While the oral examination covers normal
and emergency operating procedures, the
NRC does not directly observe operators
using these procedures. (350)
RELATED ACCIDENTS AT OTHER PLANTS
Three Mile Island was not the only nuclear fa-
cility to experience the kind of events that occurred
in the early stages of the March 28, 1979 accident. 61
The Special Investigation confirmed that the prob-
lems experienced at TMI had a parallel elsewhere
in the industry. Two accidents proved of par-
ticular interest.
THE OCONEE ACCIDENT
On June 13. 1975, a minor accident occurred at
Unit 3 of Duke Power Company's Oconee Nuclear
Generating Station in Oconee County, South Caro-
lina. This accident was quite similar to the early
stages of that at TMI-2.
81 See "Three Mile Island in Perspective : Other Nuclear Accidents," Appendix A, pp. 219ff.
76
The plant, which is equipped with a Babcock &
Wilcox reactor, was at 15 percent power and in
the process of shutting down for maintenance
when a loss of feedwater initiated a reactor trip.
(351) Pressure in the primary system increased to
the point where the pressurizer relief valve opened.
As was to happen at TMI-2. the valve failed to
close when pressure in the primary system de-
creased. Further, the valve position indicator light
in the control room malfunctioned and indicated
that the valve was closed. (352)
With pressure down, the HPI system activated.
Water continued to flow out of the stuck-open
valve into the reactor coolant drain tank. Eventu-
ally the tank ruptured, spilling approximately
1.500 gallons of reactor coolant into the contain-
ment. (353)
The operators diagnosed the leak and closed the
block valve before the water in the primary system
boiled. The situation never became serious enough
to damage the nuclear fuel. (354)
The valve failure was caused primarily by a
build-up of boron in the valve, a problem that was
later corrected. Once repaired, the valve was tested
to ensure that the position indicator functioned
properly. Duke Power management informed the
operators, following its analysis of the incident,
that closing the block valve was the proper action
in such an occurrence. (355)
XRC's Region II Office reviewed Duke Powers
analvsi>. raiswl some additional questions and ulti-
mately found the analysis to be satisfactory. (356)
The event was routinely reported in XUREG-
0020. the XRC"? "Grey Book." However, the XRC
did not perceive any generic safety significance
and did not further notify other licensees. (357)
B&W likewise reviewed the event, since its
equipment was involved, and determined that the
problem with the PORV could have generic im-
plications. As a result, all B&W plant owners were
advised to inspect the PORV's periodically."
(358)
THE DAVIS-BESSE ACCIDENT
On September 24, 1977. there was a minor ac-
cident at the Davis-Besse T*nit 1 plant operated
by Toledo Edison in Ohio. It was also verv simi-
lar to the early minutes of the Three Mile Island
accident.
While the plant was at nine percent power, feed-
water problems caused primary svstem pressure
to rise to the point at which the pressurizer relief
valve opened. Again, the valve failed to close, this
time because of a missing relay in the valve's con-
trol circuit. It opened and closed nine times within
40 seconds before sticking open. (359)
Pressure in the primary system dropped, actu-
ating the HPI system. Water flowing out through
the PORV eventually caused the reactor coolant
drain tatik to rupture, and approximately 11.000
gallons of water were released into the contain-
ment. (360)
Although at first the symptoms were common-
place water in the pressurizer initially dropped
as primary system pressure dropped an unusual
condition soon became apparent. Water in the pres-
surizer began to rise and reached a maximum level
as the coolant approached the boiling point. The
operators responded to the filling pressurizer by
turning HPI off after about four minutes, (361)
Thus, water being lost through the undiagnosed
open relief valve was not being adequately re-
plenished. Coolant in the primary system began
to boil.
The operators noticed a combination of indi-
cators, particularly high pressure in the contain-
ment and a ruptured drain tank (362) that ulti-
mately led them to diagnose the situation. They
isolated the leak after 21 minutes, before sufficient
coolant had been lost to threaten the nuclear fuel.
(363)
ANALYSIS OF THE ACCIDENT
Toledo Edison. Babcock & Wilcox. who had
supplied the reactor, and the XRC all analyzed
the accident. The reviews focused primarily on
the mechanical problems associated with the
PORV: less attention was paid to operator ac-
tions. (364)
One issue related to operator actions was iden-
tified as having potential safetv significance
premature termination of the HPI. The Office of
Xuclear Reactor Regulation. XRC, suggested to
the Office of Inspection and Enforcement at head-
quarters that it should ask that Toledo Edison ad-
dress the matter of operators turning off HPI in
its formal report. (365) However, according to
the XRC. "no significant action resulted from this
effort." (366)
The XvRC reported the incident in the Grey
Book, but did not further notifv the utilities.
(367) Xor did the inspectors' analyses reveal any
generic safety concerns. (368) in part because they
believed the problems were unique to the incident.
(369)
Some months later James Creswell, a Region
III inspector, did raise several issues. One was
that the operators might have incorrectly turned
off HPI. As a result, emergency procedures at
" It is not known at this time what caused the TMI-2 PORV to stick open, as it cannot l>e removed and inspected
until the containment can be entered.
77
Davis-Besse were modified to caution operators
against turning off HPI in the event of a leak in
the pressurizer. (370) This possibility was not rec-
ognized as a generic safety concern, and the NRC
failed to take further action or to notify other
utilities. (371)
When B&W reviewed the incident, it too con-
cluded that the circumstances at Davis-Besse had
been unique to that plant. (372) As a result, other
utilities owning B&W plants were not informed.
(373) Although several B&W engineers inde-
pendently questioned the premature termination
of HPI, internal discussions regarding a change
in operator instructions were still ongoing at the
time of the TMI-2 accident. (374)
THE MICHELSON REPORT
At about the time of the Davis-Besse accident,
TVA engineer Carlvle Michelson, 63 who was also
a consultant to the NEC's Advisory Committee on
Reactor Safeguards (ACRS) , was undertaking an
analysis of hypothetical small break loss-of-cool-
ant accidents at B&W plants. His conclusions were
contained in a draft report dated September 1977
(375) and in a revision of the report dated Jan-
uary 1978. (376) In both he described and ex-
plained how he believed the primary coolant sys-
tem would behave in the case of very small breaks.
The Michelson Report did not address actual
accidents, but rather its conclusions about cer-
tain hypothetical accidents relate to events at TMI
in March 1979.
Of particular relevance to TMI-2 was Michel-
son's analysis of the behavior of the pressurizer.
He concluded that :
A full pressurizer is not considered a re-
liable indication for prescribing certain
operator actions such as HPI pump trip.
(377)
He stated that pressurizer level would not nec-
essarily reflect the water level in the reactor vessel.
He also noted that the reactor coolant pumps
should be turned off in such situations. (378)
Michelson strongly urged that emergency pro-
cedures and operator training cover proper actions
in the event of very small break losses of coolant.
(379)
NRC RESPONSE
In earlv fall 1977, Michelson gave a handwritten
dm ft of his renort to Jesse Ebersole. a member of
ACRS. (380) In or around October 1977. Ebersole
sent a conv to Sanford Israel, a member of the
Reactor Svstems Branch within the Division of
Systems Safety, Nuclear Reactor Regulation,
NRC. (381) Israel in turn provided a copy to Ger-
ald Mazetis, also in that branch. (382) They were
the only people in the Office of Nuclear Reactor
Regulation who knew of and had copies of Michel-
soivs draft report, (383) and neither saw anything
of concern in it. (384) Israel reviewed the report
from the perspective of the small break issue and
did not find anything "new or different." (385) He
knew Ebersole was interested in loss of natural
circulation and noncondensibles, 64 but he himself
was not. (386)
In January 1978, Israel did prepare a note that
addressed the chief concern raised in Michelspn's
report that in certain circumstances pressurizer
level would not be an accurate indication of coolant
level in the reactor vessel and operators could be
misled by their instruments to turn off the Emer-
gency Core Cooling System. (387) Israel could not
recall whether he wrote this as a result of his re-
view of the report, his knowledge of the Davis-
Besse incident, questions posed by Ebersole to the
Pebble Springs applicant, or some combination of
these factors. (388)
Israel's note, dated January 10, 1978. was signed
by his supervisor, Thomas Novak, and was distrib-
uted to about 15 people within NRR. Again, there
is no evidence to show that it produced any tech-
nical interest, and no generic safety problem for
operating plants was identified. (389) Although a
related question based on Israel's note was pre-
pared in connection with an application for a new
nuclear plant which had a Westinghouse reactor,
the NRC never sent it. as plans for the reactor
were cancelled. (390)
In January 1978, Michelson provided Ebersole
with a revised tvped version of his report, but
Ebersole did not distribute it elsewhere in the NRC
until after the accident at Three Mile Island. (391 )
BABCOCK & WILCOX RESPONSE
On April 27. 1978. TVA sent Babcock & Wilcox
(B&W) a copy of the Michelson report, which it
referred to as a preliminary draft study, (392)
along with a letter asking B&W to respond to the
concerns addressed in the report.
The B&W reviewer did not see the report as
raising a substantive safety issue and assigned it
a low priority. (393) His reply, which TVA re-
ceived nine months later (it was dated January
23, 1979) did not satisfv all of Michelson's con-
cerns. Michelson therefore sent a second letter,
dated February 8. 1979, renuesting further clarifi-
cation and additional explanation.
B&W did not inform its customers of Michel-
son's concern and had not renlied to his second
letter as of March 28, 1979. (394)
M Michelson is now Director of NRC's Office of Analysis and Evaluation of Operating Data.
"A noncondensible gas. such as hydrogen in the reactor VSSP!. cannot bp converted into a liqu'd nt its existing
temperature and pressure. If a sufficient quantity becomes lodged in the piping, it will block the flow of coolant.
78
EMERGENCY RESPONSE PLANNING
Yet another factor contributing to the difficul-
ties encountered during the March 28, 1979 acci-
dent was inadequate emergency response plan-
ning by the utility, the XRC and the State of
Pennsylvania. Planning by each failed to meet the
demands of an accident of the duration and se-
verity of TMI.
EMERGENCY PLANNING: THE UTILITY
As a prerequisite for a license, the XRC re-
quires that the licensee prepare an emergency plan
that describes such things as the licensee's emer-
gency organization, employees with special quali-
fications for handling emergencies, means of mon-
itoring radioactive releases and procedures for
notification of offsite organizations. (395) The
plan must contain detailed procedures for imple-
menting emergency responses.
The TMI-2 Emergency Plan in effect on March
28. 1979 (396) classified accidents according to de-
gree of severity :
1. Local or personnel emergencies^ which in-
volve : contamination or exposure of individuals
to excessive levels of radiation and spills in work-
ing areas: flooding, fire or other conditions that
might require first aid or evacuation of buildings
or a controlled area.
2. Site emergencies, which are triggered by high
radiation readings at vent gas monitors, 65 high
radiation levels at the perimeter of the site, or a
los? of primary coolant pressure coincident with
high pressure and/or high sump level in the re-
actor building. This class required evacuation of
all affected buildings, monitoring of the perim-
eter for radiation, and notification of. among
others, the XRC regional office and the State.
3. General emergencies, which have the poten-
tial for serious radiological consequences to the
health and safety of the public. The plan listed
several conditions that required declaration of a
general emergency, all based on radiation levels
either inside the containment building or in at-
mospheric or liquid effluents. Again, the licensee
was to notify the State and the XRC regional of-
fice as well as other agencies. In addition, it was
to initiate offsite monitoring 66 and establish an
Emergencv Control Station as soon as possible.
The XRC required that the utility meet cer-
tain obligations related to State emergency re-
sponse as a condition of issuing a license. Fore-
most was that the utility provide the State with
information throughout the accident as it was pro-
gressing. However, the NRC did not specify what
information was to be transmitted. (397)
EMERGENCY PLANNING: THE NRC
In 1979, the XRC had a number of plans, pro-
gram documents, studies and procedures covering
emergency response. (398) The basic document
describing the agency's overall goal for emergency
response was a chapter in the XRC Manual. (399)
As stated there, the goal was :
... to assure that proper actions are
taken to protect health and safety, the en-
vironment, and property from the conse-
quences of incidents which occur as a
result of XRC-licensed activities : to pro-
vide, as appropriate, for common defense
and security : and to assure that the public
is kept informed of actual or potential
hazards to health and safety arising from
such incidents. (400)
To meet this goal, five basic program objectives
were set forth:
. . . gathering or providing information,
evaluating response, coordinating with
other agencies, assisting where appropri-
ate, or directing where necessary. (401)
The chapter outlined a program according to
which, during a nuclear incident, the agency was
to set up an incident response organization con-
sisting of three basic groups: an Incident Re-
sponse Action Coordination Team (TRACT), an
Executive Management Team (EMT) and the
Commissioners. IRACT and the EMT were given
specific duties. (402) The Commissioners, on the
other hand, were vaguely charged with respon-
sibility for providing "general policy which de-
termines the overall course of action XRC takes
in response to incidents." (403) Unlike IRACT
and the EMT. the Commissioners were neither as-
signed specific duties nor charged with develop-
ing specific procedures governing their emergency
response.
Xeither the XRC Manual nor any of the other
XRC documents relating to the agency's incident
response program contained any provision spe-
cifically relating to recommendations on evacua-
tion or other protective action. They did not set
forth any role in this respect for the Commission-
ers or any other entity in the agency's incident re-
sponse organization, despite the program's stated
85 These are radiation detectors located In the plant's vent sas stacks.
K Offsite monitoring refers to the dispatching of survey teams equipped with radiation monitors to determine the
amount of radiation at various locations outside the plant's boundaries.
79
goal of protecting the health and safety of the
public. (404)
As noted, the NRC Manual outlined specific
duties for TRACT and the EMT. It delegated to
the Director of the Office of Inspection and En-
forcement (I&E) responsibility for developing
and maintaining procedures for implementing
those duties according to the type of incident;
other NRC offices were to review and approve
those procedures. (405) The Director of I&E in
turn assigned that responsibility to the Divisions
within I&E.
The NRC Headquarters Incident Response Plan
contained both the general provisions of the Man-
ual on implementing IRACT's and EMT's duties,
as well as the separate implementing procedures
prepared by the I&E divisions. 67
The I&E Office also had a Manual. One chapter
established "policy and procedures," whereas two
others provided "instructions" concerning the
agency's incident response program. All the chap-
ters in the I&E Manual predated the NRC Manual
by several years ; none had been revised since De-
cember 11, 1975. (409) As a result, in March 1979
the I&E and NRC Manuals provided for different
incident response organizations and responsibili-
ties. The failure to revise the I&E Manual sug-
gests that the NRC considered emergency response
planning a low priority. Further, it contributed to
lack of coordination among the offices that would
respond to an emergency. That lack of coordina-
tion, would become very evident during the March
28 accident.
Finally, the regional office had a plan. Although
it had been updated in February 1979, it did not
reflect the most current planning by headquarters.
For example, the plan called for the Regional Of-
fice to be the lead unit in the NRC's emergency re-
sponse, but did not define how the Region was to
interact with headquarters. The NRC Manual, on
the other hand, called for an integrated, agency-
wide response.
THE INCIDENT RESPONSE CENTER
According to the NRC Manual, the Headquar-
ters Incident Response Plan and the I&E Manual,
(410) one of the first actions to be taken in the
event of a nuclear accident was to set up an Inci-
dent Response Center. It was to be comprised of
two principal units TRACT and the EMT
which were to be located in adjoining offices in one
of the buildings at NRC headquarters in Bethesda,
Maryland. Personnel from I&E and other offices
at headquarters would be called on as support
staff and would work out of the Incident Response
Center. There was some contradiction within the
various documents over the composition of the sup-
port staff, as described later.
The Center was to be the heart of the NRC's re-
sponse to a nuclear accident. IRACT was to re-
ceive and evaluate incoming information, identify
real or potential problems and develop alterna-
tive solutions. (411) Once TRACT had received
and evaluated the information, it was to pass it
on to the EMT. TRACT was to filter and process
the incoming information for transmission to the
EMT according to guidelines spelled out in the
NRC Headquarters Incident Response Plan.
(412)
Although the plan assigned IRACT respon-
sibility for providing the EMT with "adequate"
information for EMT's decisionmaking and other
functions, it did not define "adequate" precisely.
It said only that "EMT should be provided with
evaluation of information acquired, not with de-
tails external to the evaluation, e.g., unevaluated
raw data." (413)
Both the NRC Headquarters Incident Response
Plan and IRACT implementing procedures were
specific as to how information was to flow between
TRACT and the EMT. As spelled out in the Plan,
all information was to pass through an TRACT/
EMT Liaison Officer : (414)
1. The Liaison Officer person allv comes
from the Operations Room (IRACT) to the,
Executive room (EMT) for each briefing.
2. The initial portion of each briefing con-
sists of a brief, concise statement of the situa-
tion or update of the situation.
3. After the update of the situation, the
Liaison Officer states :
The principal questions now being pur-
sued bv IRACT are ... (a concise list-
ing of those questions being pursued by
:IACT including previously submitted
EMT questions, if any.)
IRACT
IRACT was to support the decisionmaking and
policy-setting functions of the EMT and the Com-
missioners. (415) This structure was established
_ developed and approved by three of the four I&E divisions whose division
(406) Depending on the type of incident, one of the directors would become the IRACT
Trh e '^P 1 ^^^ Procedures for that division would pertain. The procedures for incidents at operating
November 29, 197? (407) devel P d b * I&B 8 Division of Reactor Operations Inspection (ROI) ; they were dated
' ROI TOrOf'OfJ 111*0*3 Pflllpfl ff\T TT? A f""P nn/1 fe a 4-4- tV 4. v. i
i nl^ mm !^ niC f t0r A (5) ^ lant Systems Effects group, (6) Radiological and Environmental Effects group,
Response Center Operations Staff. (408)
80
in 1978. 68 The specific actions to be undertaken by
TRACT and its support staff were spelled out in
the TRACT implementing procedures, (419)
within the framework of the general guidelines
contained in the XRC Manual. The Manual stated
that the procedures were to be "designed to imple-
ment the Incident Response Program . . . with a
minimum of confusion." (420) It called, among
other things, for procedures for "identifying and
assembling IRACT support staff" and for "issuing
oral or written directives to licensees." and, in gen-
eral, for whatever "other procedures [were]
deemed necessary to meet incident response ob-
jectives/' (421)
The Manual also specified that IRACT was to
be composed of four I&E Division Directors, two
Division Directors from the Office of Xuclear
Material Safety and Safeguards (XMSS) 69 and
one Division Director from the Office of Xuclear
Reactor Regulation (XRR). (423) The IRACT
Director was to be selected from among the four
I&E Division Directors, and the Director of I&E
was to be a member of the EMT. (424) The team
was to be assisted by a support staff drawn from
appropriate XRC offices, to be determined on the
basis of the type of accident.
The XRC Manual stated that XRR was to pro-
vide IRACT a team member and support staff and
that these people were to have important roles. 70
Xeither the XRC Headquarters Incident Re-
sponse Plan nor the ROI implementing proce-
dures, however, specified any procedures by which
the XRR representatives were to carry out their
functions, although the Manual directed that those
documents contain that information. This was an-
other example of the incompleteness and lack of
coordination in the XRC's emergency planning.
The Executive Management Team
The XRC Manual provided for an Executive
Management Team (EMT), to be composed of the
XRC's Executive Director for Operations and the
Directors of I&E, XRR and XMSS or their desig-
nated alternates. (426) The EMT's role was to
"transform Commission policy into specific guid-
ance for the response organization and make major
decisions affecting XRC's response actions." (427)
EMT's responsibilities and duties were fur-
ther delineated by the Headquarters Incident Re-
sponse Plan. It stated that the EMT would have
to decide such issues as "should XRC provide as-
sistance or on-site direction?" (428) The EMT
was expected to approve "specific NRC direc-
tives to the licensee during incident response."
(429) In addition, the EMT was charged with
notifying "senior governmental officials," includ-
ing the White House and the Chairman of the
XRC, of the incident. Another EMT duty was to
coordinate "XRC offices' joint activities related to
the incident" and "policy with other agencies."
(430)
The Commissioners
The XRC Manual defined the Commissioners'
responsibility as one of providing general policy
on the XRC's overall emergency response. (431)
That policy would provide the EMT with the
framework for managing the incident response
organization. The EMT would transform that pol-
icy into "specific guidance" to IRACT and the rest
of the XRC's response organization.
According to the Headquarters Incident Re-
sponse Plan, the IRACT Communications Officer
was to notify the Commissioners, except for the
Chairman, of an emergency. EMT was to notify
the Chairman. (432) The Communications Officer
also was to update any Commissioners outside the
response center on an accident's evolution. (433)
The plan did not require that the Commissioners
be stationed in any specific location or even that
they deliberate as a body.
88 Prior to the formulation and approval of XRC Manual Chapter 0502 in February 1978, neither the Commissioners
nor the senior technical staff outside I&E had a defined management role within the agency's incident response program.
Commission policy, as reflected in I&E's Manual Chapter 1300, was simply that the "actions taken in response to incidents
will be planned and coordinated" by IRACT. No mention was made of an EMT. and executive-level oflScials and the
Commissioners were referred to only in the context of being kept informed of actions that IRACT might undertake. (416)
IRACT ha:! consisted solely of high-level I&E officials, with the Director of I&E designated as the Director of IRACT.
(417) Officials from other XRC offices were responsible only for contributing to the IRACT support staff "when
necessary." (418)
Adoption of XRC Manual Chapter 0502 and continuing revision of the XRC Headquarters Incident Response Plan
led to significant changes in the structure of the agency's response program. The composition and scope of IRACT's
responsibilities and authority were changed. More detailed guidance was given on how IRACT was to implement its part
of the agency's response and on the relationship of IRACT with the Commissioners and the newly established Executive
Management Team.
* The XRC Manual and Headquarters Incident Response Plan conflicted over XMSS participation in the incident
response program. The Manual called for XMSS participation on both IRACT and the EMT, while the Plan made XMSS
participation on IRACT dependent on the type of accident, calling for the "appropriate XRR or XMSS Division Director"
to become the fifth IRACT member. Under the provisions of both the plan and the Manual, XMSS was to participate on
the EMT. (422)
T The XRR participant on the IRACT support staff "evaluates information with respect to the likely future course of
the incident"; "evaluates corrective action taken and proposed by reactor licensees in response to [the] incident":
"determines alternate courses of future action available" : "evaluates the feasibility of assistance to the licensee or
others, recommends to the IRACT the initiation of such assistance, and participates in the provisions of assistance as
appropriate" ; and "evaluates the need for formal intervention by XRC and recommends the initiation of such intervention
to the IRACT." (425)
81
The various plans did not call for the Commis-
sioners to take an active role in the NRC's emer-
gency response. Commissioner Ahearne described
the Commissioners' view of their role :
As far as the issue of what is the role of
a Commissioner . . . during emergency
response, mv understanding of it prior to
and certainly during [the accident] was
that the way the NEC system was de-
signed was for the senior technical peo-
ple in the agency to be responsible for
monitoring and taking whatever action
might be necessary as far as the technical
issues. (434)
Commissioner Gilinsky noted :
. . . generally speaking, the technical,
minute-by-minute decisions and recom-
mendations have to be handled by our
staff. And the Commissioners have got to
deal with things that are more general
in nature . . . but the technical questions
have got to be examined bv the staff, and
it is they who have to be in direct touch
with the licensee as well as counterparts
in the State. (435)
Chairman Hendrie and Commissioner Ahearne
explained that the NRC Manual assigned the
Commissioners only a policy-making role based
on the premise that accidents would be over very
quickly. For this reason, "the Commissioners
themselves were not assumed to have a role in par-
ticipating" in the agencv's response, and the re-
sponse organization would make "whatever deci-
sions had to be made." (436) The Commissioners
were envisioned "as sort of an ultimate policy deci-
sionmaking -body for the agency for those things
that might follow in the aftermath" of an accident.
(437)
Commissioner Gilinsky pointed out. however,
that the Commissioners should be prepared to be
flexible :
It is hard to put down in a manual a
set of rules that will cover every possi-
bility. It is the nature of accidents that
unusual things turn up and often re-
quire unusual solutions. The Commission-
ers are in charge of this agency, and ulti-
mately have to be responsible for what it
does. And it may be that decisions will be
required of them that thev didn't expect
to have to make . . . and they have to be
ready to do that (438)
Commissioner Gilinsky's observations reflect les-
sons learned from the Commissioner's response on
March 28. On that day there were several points at
which a well-informed, actively involved Commis-
sion might have made important contributions.
The foremost example involved the need for pro-
tective action. During the Subcommittee's hear-
ings following the accident, three Commission-
ers indicated that they would have considered pro-
tective action on March 28 had they had the
information available to the utility. 71
NRC REGION I
The various plans also called for the activation
of a Regional Incident Response Center. The Re-
gion I Incident Response Plan designated the rear
half of the main conference room at the Region I
offices in King of Prussia, Pennsylvania, as the
location for the center. (439) Two teams were to be
set up: (1) a Regional Incident Response Action
Coordination Team and (2) an Onsite Inspection
Team. (440)
When notified of an incident at a nuclear facility
under its jurisdiction, the Region was to classify
it according to severity and decide whether to ac-
tivate the response center and dispatch an inspec-
tion team to the facility. It also was to notify NRC
Headquarters and other incident response support
organizations. (441)
Further defining Region I's response were the
implementing procedures in the Region I Incident
Response Plan. 72
While a Regional Office might have the lead re-
sponsibility in the early stages of an accident, as
an arm of I&E it was ultimately subordinate to
IRACT at headquarters. (443)
NRC EMERGENCY COMMUNICATIONS
On March 22. 1975, a major fire broke out at the
Browns Ferry nuclear powerplant in Decatur,
Alabama. It took hours to bring the reactor under
control. The NRC had substantial difficulty in re-
sponding effectively, particularly becaiise of weak-
nesses in its communications system.
After the accident, the NRC appointed a Special
Review Group to "distill from the available infor-
mation those lessons that should be learned for the
future." (444)
By February 1976, the NRC's Special Review
Group had analyzed the agencv's response. In its
report, it described the flow of information during'
the accident from plant operators to onsite NRC
inspectors to the regional office to NRC headquar-
ters and on to other government officials. (445)
The Group commented : "The well-known game of
'password' shows how poorly information is trans-
mitted through such chains." (446)
See "The Accident at Three Mile Island : The First Day," pp. 150-151.
"This document was revised in February 1979. (442)
82
The Group recommended that communications
facilities (which is left unspecified) be provided
and that "the problem deserves a deeper study and
more expertise than [we] are able to bring to bear
on it. and that a systems study (who should com-
municate with whom, when and how?)" be com-
missioned. (447).
In June 1976. the NRC hired the MITRE Cor-
poration, a consulting firm, to conduct a study on
"Communications and Control to Support Incident
Management." MITRE was "to define new com-
munication concepts, requirements and procedures
which will allow the Nuclear Regulatory Commis-
sion to respond more effectively to nuclear inci-
dents involving its licensees." (448) The original
contract was for $94.000.
MITRE issued a two-volume report in Novem-
ber 1977. (449) It outlined three possible communi-
cations system?, based on different roles the NRC
might assume in responding to accidents. For each
alternative, the study provided startup procedures.
requirements for making the system operational on
an interim basis, and the actions necessary to reach
full operational capability. (450)
The concept behind the first system was that the
NRC would simply monitor the course of a nuclear
incident :
In this concept the NRC's involvement
would be limited to monitoring the activi-
ties of the various response units and co-
ordinating Federal information exchange.
(451)
In this case, the NRC would depend on other
organizations for information.
The second alternative conceptualized the NRC
as an advisor to the licensee, but dependent on it
for information :
This concept would allow the [Incident
Response Center] to provide detailed ad-
vice, if needed, based on information sup-
plied by the calling party or on file in the
[Center]. (452)
The third option was particularly noteworthy
in that, in many respects, it foreshadowed how the
NRC would come to see itself and how it would
restructure its incident response program follow-
ing the TMI accident. (453)
The third system envisioned an NRC that would
serve as an advisor to its licensees on the basis of
data on the status of the reactor that the NRC
would collect independently :
In this concept, the [Incident Response
Center] would receive sensor information
transmitted directlv from reactor instru-
mentation. Transmission would probably
be triggered by [automatic] alarm. Dur-
ing normal operations the [IRC] could
dial up any reactor on a standard tele-
phone line to scan the reactor instrumen-
tation data. In an incident, the alarm
would trigger automatic dial-up from the
reactor site to the [IRC] where data
would be recorded. The [IRC] would also
be able to select any number of the sensor
inputs for concentrated attention.
By adding a source of reactor perform-
ance data independent of licensee person-
nel, the [IRC] may be able to help an-
ticipate new complications in an inci-
dent and to offer the [offsite response cen-
ter] alternative remedies. The licensee
would still decide, ultimately, what in-
structions to pass on to his site personnel.
The capability to assess the situation in-
dependently, however, provides the
[IRC] the information base required to
intervene in the licensee response if it
should ever be necessary. (454)
At the time of the accident, the NRC had ex-
pressed its intention to implement the third alter-
native but had not yet established the necessary
communications system. In the interim, it adopted
the second alternative advisor, dependent upon
the licensee for data. 73 (455) It did so despite a
prophetic warning from the consultant:
The dependence on information fur-
nished by the calling party [licensee] or
on file in the [Incident Response Center]
is the most obvious limitation [of this
option], since the [Center] is unlikely to
have enough information to anticipate a
problem not already noted bv the caller.
(456)
Lee Gossick, a member of the EMT. explained
to the Special Investigation staff the assumption
underlying incident response at. the time. It pro-
vided a possible rationale for selection of the sec-
ond option. The assumption was that an event
would last onlv a short time. The emergency re-
sponse drills of the Incident Response Center prior
to the accident were based on that assumption and
did not, provide experience with accidents of long
duration. (457) "The Three Mile Island thing was
an event unlike that which any of us ... antici-
pated," Gossick told the Special Investigation
staff. (458).
Edson Case, like Gossick a member of the EMT
on March 28. confirmed that the drills were based
"On March 28 the XRC was dependent on the utility for data on key plant conditions, such as hotlee temperatures,
incore thermocouple readings, and the status of natural circulation. For most of the first day. it received incomplete or
erronpous information or was unable to Ret answers to requests. See "The Accident at Three Mile Island : The First Day,"
pp. 110-111, 119-121, 126-128, 131-132, 137-138. 143-145.
83
on events of short duration. (459) He said that
planning for accident scenarios was based on a
range of accidents, including some that would in-
volve large releases of radiation, but that generally
the accident would be over before the NRC could
play an active role. (460) According to Case, the
NRC conceived its role to be one of directing off site
actions to minimize public exposure to radiation,
rather than giving advice or direction to the li-
censee on how to operate his plant. (461)
It should be noted that Case's perception of the
NRC's role conflicts with that spelled out in both
the NRC Manual and the alternative the NRC
chose from among the three provided by the
consultant.
The week of March 28 at Three Mile Island
underscored dramatically the inaccuracy of the
presumption that accidents would be of short
duration. 7 *
EMERGENCY PLANNING: THE STATE
The NRC had, and has, no regulatory authority
over a State's emergency response or plans. (462)
Therefore, there was no requirement that the
State in which a nuclear plant was located have an
adequate emergency response plan. (463) Nor was
there any requirement that an adequate State plan
be in existence before a nuclear power plant lo-
cated in that State would be licensed.
The Commission had general authority to im-
pose such a requirement, if it determined that such
a requirement was necessary to protect the public
health and safety. The Commission never made
such a determination.
The States could, however, voluntarily submit
their emergency plans to the NRC for "concur-
rence." As of March 28, 1979, eleven States had
secured NRC concurrence; Pennsylvania was not
among them.
The NRC's regulations recognized certain State
responsibilities, the most important of which is to
decide on protective action such as evacuation. In
turn, local governments, with State support, would
implement that decision. (464)
EMERGENCY MANAGEMENT
In Pennsylvania, the designated lead agency
for the State's response in the event of an emer-
gency at a nuclear plant was the Pennsylvania
Emergency Management Agency (PEMA). Its
role was to assure prompt, proper and effective
discharge of basic Commonwealth responsibilities
related to civil defense and disaster preparedness,
operations, and recovery." (465) A Council headed
by a Chairman was to set the Agency's overall
policy. A State director hired by the Council su-
pervised PEMA's activities. (466)
PEMA headquarters were to be located in the
basement of the Transportation and Safety Build-
ing. During an accident, the Emergency Opera-
tions Center (EOC) was to be located there.
PEMA was to enlist and coordinate the assistance
of other State and Federal agencies as the situa-
tion required. Upon activation of the EOC, af-
fected State agencies were to dispatch representa-
tives to cubicles within the Center. In the event
that protective action became necessary, PEMA
was to be responsible for its implementation.
The Bureau of Radiation Protection (BRP), a
division of the Pennsylvania Department of En-
vironmental Resources (DER), is an important
component of Pennsylvania's emergency response
organization. Its Division of Environmental
Radiation is routinely involved with environ-
mental surveillance, laboratory activities and
emergency planning. (467) During an accident
involving releases of radiation to the environment
that could require protective action, BRP was to
serve as PEMA's technical advisor. In fact, once
PEMA had been notified of an emergency at a
fixed nuclear site, it would no longer talk directly
to the site, but would rely on BRP personnel. BRP
would receive information from the site, coordinate
radiation monitoring and advise the Common-
wealth on protective action such as evacuation.
Although PEMA had divided the Common-
wealth into several areas, each with its own small
office, the political subdivisions were to carry out
protective action and other tasks as required. (468)
During an emergency, county and local emergency
preparedness directors were to marshal personnel
and equipment from county and municipal agen-
cies. They were to receive information from and
be coordinated by PEMA operations personnel.
STATE EMERGENCY PLANS
The State had three emergency plans that out-
lined its response ; these plans were distinct from
the Met Ed plan, described above. PEMA had a
Departmental Operations Plan that served as the
general emergency guide for the Commonwealth
of Pennsylvania. The August 1978 edition of An-
nex E of the PEMA plan dealt with radiological
incidents at fixed nuclear sites in Pennsylvania.
BRP had two response plans which applied to
TMI. The first was a general plan which applied
to all nuclear plants in the Commonwealth. (469)
The second was limited to TMI. (470) While noti-
fication channels were similar to those in Met Ed's
74 There is evidence that this presumption contributed to the communications difficulties the NRC had on the first day.
See, in particular, "The Accident at Three Mile Island : The First Day," pp. 120-121.
84
site emergency plan, the classifications of nuclear
incidents were different. 75
As discussed above, none of the plans had been
submitted to the XRC for voluntary review and
concurrence. (477 j
In addition to these plans, county directors were
to have written umbrella plans, along with annexes
that they were to submit to PEMA for approval.
< 47-> ) However, local Civil Defense personnel were
usually volunteers, and many had no written
emergency plans at the time of the accident.
EVACUATION
A crucial question that utilities, the XRC and the
States have to address in the event of a nuclear
accident is whether to take protective action, and
most particularly, whether evacuation is necessary.
Since January 1973. the Environmenal Protec-
tion Agency (EPA) has had responsibility for
ng Federal and State guidelines governing
protective action in relation to actual or projected
releases of radioactivity beyond the boundaries of
XRC-regulated facilit'ies. "(479) The guidelines
cover levels of radiation at which protective
action is mandatory, and methods for projecting
dose rates so that a determination of the need for
protective action can be made in advance of actual
releases.
As described below, the version of the EPA
guidelines in effect in March 1979 was incomplete
with respect to projecting dose rates: it did not
spell out clearly the criteria to \te used in making
the appropriate calculations. Most important was
the failure to define "plant conditions*' and how
they were to be used, particularly in terms of wor-
sening conditions, to project releases and dose
rates.
EPA GUIDELINES
In September 1975. the EPA established cri-
teria for determining the need for protective
action, such as evacuation, in response to nuclear
accidents that could expose the public to radia-
tion. These criteria were set forth in the EPA's
"Manual of Protective Action Guides and Protec-
tive Actions for Xuclear Incidents."' 76 (481)
A key chapter of the EPA Manual Chapter 5.
"Application of Protective Action Guides for Ex-
posure to Airborne Radioactive Materials from an
Accident at a Xuclear Power Facility" was being
revised at the time of the accident. A draft of this
chapter was ready, in January 1979. but it was not
issued until June 1979, (482) several months after
the accident.
Other chapters in the 1975 Manual detailed a
number of steps in protective action decision-
making. Beyond pre-accident planning, they
were: (1) evaluation by the facility operator of
the projected effect of a nuclear accident on. pub-
lic health and safety, (2) notification by the utility
of State and local officials that an accident had
occurred, and (.3) collection of additional informa-
tion and/or warnings to the public. (483) Even if
an initial determination was made that protective
action was not warranted, the State still would
need to collect and evaluate information hi order
to assess whether that determination should be
modified. (484)
The facility operator was to provide "detailed
information.'' such as projected doses, to the public
and to the State. These doses were to be estimated
from data obtained at the point of release or from
"releases anticipated for particular types of nu-
clear incidents." (485) If the operator did not
provide that information,
... the emergency plans of the State
should provide for action in the immedi-
ate downwind area of the facility based
on notification that a substantial release
has occurred or that plant conditions are
such that a substantial release potential
exists. (486)
As was evident on March 28. neither the util-
ity, the XRC nor the State were clear what data
"The General Procedures and Guidelines Manual of the Bureau of Radiation Protection (471) has a classification
:n for accidents based on XRC regulations. These class?s can be briefly described as :
Class I Incidents with no radiological consequences, but of potential public interest.
Class II Abnormal occurrence, i.e.. major reduction in protection for health and safety.
Class HI Threat of radioactivity offsite. e.g.. LOCA. (472 i
These classes are the same as those in PEMA's Annex E. (473)
However, the BRP site specific plan for TMI (474) has four types of accidents :
Type 1 Unplanned release to Susqnehanna River.
Type 2 Potential release to the atmosphere.
Type 3 Release to the atmosphere as a result of system failure.
Type 4 Maior failure with failed safeguards. (47." i
The Three Mile Island nuclear station site emergency plan has yet another system of classification, each with its own
descriptions and notification procedures. These classes are :
< 1 1 local emergencies,
i 2 i site emergencies, and
181 general emergencies. (476)
The inconsistent and overlapping classifications of nuclear accidents contained in the various plans reveal little
attempt at uniformity.
'' Protective Action Guides (PAGs) describe "projected radiological doses ... to individuals in the general
population that warranted protective action following a release of radioactive material." (480)
the utility was to transmit, and the utility did not
provide the State with information on plant con-
ditions. 77 In effect neither the utility nor the NRC
considered uncertainty as to uncovering of the
core a condition that warranted serious consider-
ation of evacuation. 78
REVISED EPA GUIDELINES
The revisions of Chapter 5 referred to above
elaborate to a limited extent on the "detailed in-
formation" needed by the State. The State is to
have (1) Protective Action Guides (PAGs) ad-
justed for local conditions and (2) projected doses
for comparison with the adjusted PAGs. (487)
The projected doses were to be derived from one or
more of three sources :
(1) plant conditions.
(2) release rates and meteorological condi-
tions, or
(3) offsite radiological measurements, or
combinations thereof . (488)
An appendix to the Manual, dated January
1979, defines the first of these three data bases as
"reactor system status." However, it provides no
further guidance as to how this information is to
be used in connection with protective action.
Rather, it notes : "Dose projection based on reac-
tor system status will be primarily the responsi-
bility of nuclear facility officials and will not be
discussed here." (489)
The updated version also assumed that the fa-
cility operator is the most likely to have accurate
information on plant conditions. As Floyd Galpin,
EPA's Director of the Division of Environmental
Analysis, wrote the NRC following the accident :
. . . [A] 11 of our guidance to States have
implied a first order dependence on the
facility operator for information on the
releases. . . . (490)
PROJECTING DOSE RATES
Of the three sources of data, field measurements
were considered the most accurate for making pro-
jections, as they reflected dose rates at the time of
measurement. Continuous monitoring would pro-
vide data that the State could use to evaluate initial
and subsequent protective action decisions. (491)
This source, however, has one major weakness.
It assumes there will be no change at the site that
could abruptly alter the release rate. This assump-
tion might not bo correct in the case of an ongoing
incident. For that, the first source of data specified
in the January 1979 Appendix accurate and up-
to-date information on plant conditions (reactor
system status) would be needed. That would in-
clude actual or anticipated changes in the condi-
tion of key components or systems. Only with this
information can future releases be projected effec-
tively. (492)
Thus "plant conditions" or "reactor system
status" are key elements in projecting dose rates,
on the basis of which the need for protective action
can be determined. The revised version of the
Guidelines, which neither the State nor the utility
had seen in 1979, does not go far beyond the older
version of the EPA Manual, which provided in-
sufficient guidance. It states only that plant condi-
tions are defined as "reactor system status" and
should be used in determining projected doses in
consideration of protective action.
The EPA, in response to a request by the Sub-
committee, stated that by "plant conditions" it
means "observable parameters onsite that could bo,
used to predict, the course of the accident, includ-
ing its seriousness with regard to releases." (493)
The EPA also told Special Investigation staff that
the operator is (and was) to report to the State
whether specific safeguards might fail and what
the consequences would be, what offsite releases
in what ranges would follow, what checks were in
place and the time before an event might take
place. Public health officials could then take ad-
vantage of maximum lead times. (494) The State,
of course, must have people who can understand
the dose projections.
This definition of plant conditions as plant pa-
rameters was not specified in the Manual, nor does
the Manual define which of the hundreds of pa-
rameters should be considered. Further, it does not
provide guidance as to what should be done if the
reliability of a key indicator is in doubt, for ex-
ample, water level in the core.
Beyond this, neither the revised nor the old
Manual spells out adequately who is responsible
for protective action, and neither version specifies
any role for the NRC, despite the NRC's mandate
to protect the health and safety of the public.
" See "The Accident at Three Mile Island : The First Day," pp. 135-136.
" See "The Accident at Three Mile Island : The First Day," pp. 134-135.
86
Chapter 7
Accident At Three Mile Island:
The First Day
87
Control room personnel discussing plant conditions and strategy during the accident
88
PRINCIPAL PARTICIPANTS IN THE ACCIDENT |
AT THE THREE MILE ISLAND PLANT- |
AHEARNE, John F. NRC Commissioner. One of three who spent part of first day of accident at
the Incident Response Center in Bethesda. Named Acting Chairman of the NRC in November 1979.
Told by Edson Case at 9 a.m. core probably had been uncovered.
ARNOLD, Robert. Vice President for Generation of the GPU Service Corporation. Said he
questioned control room personnel on core uncovering. Contributed to the strategy that finally
succeeded in returning the plant to stable conditions in late afternoon.
BENNETT, Skip. Instrumentation Foreman at TMI. Deduced, based on incore thermocouple
readings, that core had been uncovered.
BENSON, Michael L. Lead Nuclear Engineer at TMI-2. Arrived at the Unit 2 control room
about 7 a.m. Deduced that neutron detectors showed excess neutron leakage from core.
BRADFORD, Peter. NRC Commissioner. One of three who spent part of first day at the
Incident Response Center.
BRUNNER, Eldon. NRC Branch Chief at Region I. First official at Region I to receive word
of the incident. Activated Regional Incident Response Center.
BRYAN, Ken. Shift Supervisor at TMI-1. Arrived in Unit 2 control room eight minutes into
the accident.
CASE, Edson. Deputy Director of NRC's Office of Nuclear Reactor Regulation. Was member
of the NRC Executive Management Team. Advised Commissioner Ahearne at 9 a.m. that core might
be uncovered.
CHWASTYK, Joseph. Shift Supervisor at TMI. Only person whose statements indicate he
correctly attributed pressure spike in the reactor building to a hydrogen burn.
CRAWFORD, Howard C. Nuclear Engineer at TMI. Performed initial projected dose-rate cal-
culations about 7 :15 a.m. based on containment dome-monitor readings.
CRITCHLOW, PauL Governor Thornburgh's Press Secretary and Director of Communications.
Was involved with press statements, news conferences and briefings conducted by Pennslyvania
State officials.
DAVIS, John. Acting Director of NRC's Office of Inspection and Enforcement. Member of NRC
Executive Management Team. Activated the Incident Response Center in Bethesda on March 28.
DENTON, Harold R. NRC's Director of Nuclear Reactor Regulation. Became member of the
Executive Management Team in the afternoon.
DORNSIFE, William P. Nuclear Engineer with the Pennsylvania Department of Environ-
mental Resources, Bureau of Radiation Protection. Only nuclear engineer with the State emergency
response organization. Was on-call duty officer on March 28.
DUBIEL, Richard W. TMI-2 Supervisor of Radiation Protection and Chemistry at TMI-2.
In charge of radiation-protection activities, including assessment of onsite and offsite monitoring
during accident.
EISENHUT, Darrell. Deputy Director of NRC's Division of Operating Reactors. Assembled
reactor-systems and radiological-assessment teams for NRR. Periodically briefed Harold Denton and
relayed information between Babcock & Wilcox and Victor Stello.
FAUST, Craig. Control Room Operator at TMI-2. Present in control room when accident be-
gan. One of four responsible for initial response to accident.
1 Unless otherwise indicated, descriptions refer to positions held or roles played on Wednesday, March 28, 1979.
89
51-058 0-80-7
PRINCIPAL PARTICIPANTS IN THE ACCIDENT
FLINT, John. Babcock & Wilcox, Engineer and Start-up Representative at TMI-2. Arrived
Unit 2 control room about 9 a.m. Was among first to recognize core had been uncovered and super-
heated conditions in the reactor vessel.
FOUCHARD, Joseph. Director, NEC's Office of Public Affairs. Responsible for generating
press releases issued by NRC from the EMT office during the accident.
FREDERICK, Edward. TMI-2 Control Room Operator. Present in the control room when
accident began. One of four responsible for initial response to accident.
GILBERT, Bob. Instrumentation Technician at TMI-2. Arrived in cable room while incore
thermocouple readings being taken. Did not interpret readings to indicate core had been uncovered.
GILINSKY, Victor. NRC Commissioner. Acting Chairman of NRC first day while Chairman
Hendrie away. At NRC headquarters in Washington, D.C., most of day. Told by Stello at 4 :30 p.m.
that core was uncovered.
GOSSICK, Lee V. Executive Director for Operations at NRC. Director of NRC Executive
Management Team. Participated in conference call at 4 :30 p.m. to Commissioner Gilinsky concern-
ing uncovering of the core.
GRIER, Boyce. Director of NRC's Region I. Notified John Davis at NRC headquarters of
accident. Coordinated early Region I response with Smith and Brunner.
GRIMES, Brian. Assistant Director of Engineering and Projects in NRC's Office of Nuclear
Reactor Regulation. Office representative on support staff of Incident Response Center in Bethesda.
HAVERKAMP, Donald R. Project Inspector in NRC's Region I. Served as liaison on the first
day of the accident between Region I and the site.
HENDRIE, Joseph. Chairman of the NRC. Was absent first day of the accident.
HERBEIN, John G. Met Ed's Vice President for Nuclear Generation. Became utility spokes-
man to the press, the Lt. Governor, and the NRC. Contributed to strategy that finally brought plant
to stable conditions in late afternoon.
HIGGINS, James C. Inspector at NRC's Region I. Member NRC onsite inspection team. One
of the two NRC inspectors in Unit 2 control room. Said he was unaware of pressure spike in the
reactor building and did not report it to NRC.
HITZ, Gregory. Shift Supervisor at TMI. Served as intermediary between Victor Stello and
the TMI-2 operators. First at site to learn of Stello's concerns regarding superheated steam and
uncovering of the core.
KENNEDY, Richard T. NRC Commissioner. Notified of accident by John Davis at 8 :52 a.m.
Spent day at NRC headquarters in Washington, D.C.
KISTER, Harold. Inspector at NRC's Region I. Manned the phones to TMI and to IRACT.
Received Victor Stello's request around noon for incore thermocouple readings.
KUNDER, George. Superintendent of Technical Support and the on-call Duty Officer at
TMI-2 during morning of first day. Placed in charge of technical support and communications.
LOGAN, Joseph B. Superintendent at TMI-2. Charged with ensuring that all required proce-
dures and plans were reviewed and followed.
MEHLER, Brain. Shift Supervisor at TMI. Arrived in TMI-2 control room about 6 a.m.
Recognized PORV was stuck open and ordered block valve closed at 6 :22 a.m. Deduced steam in the
hotlegs around same time.
MILLER, Gary. Station Superintendent at TMI. Arrived TMI-2 control room shortly after
7 a.m. Became Director of Met Ed's Emergency Command Team. Said he was unaware core had
been uncovered and said he did not know about the pressure spike in reactor building.
90
AT THE THREE MILE ISLAND PLANT
MOSELEY, Norman. Director of NRC's Division of Reactor Operations Inspection. Was
Director of the Incident Response Action Coordination Team. At Victor Stello's direction, raised
issue of superheated steam with James Higgins in Unit 2 control room at 4 :30 p.m.
PORTER, Ivan. Met Ed's Lead Instrumentation Engineer. Collected incore thermocouple
readings around 8 a.m. and told Gary Miller they were unreliable. Oversaw installation of resistance
bridge for reading hotleg temperatures.
ROGERS, Leland. Babcock & Wilcox's Site Operations Manager at TMI. Was in TMI-2 control
room much of day and served as liaison with B&W's Division of Nuclear Generation in Lynchburg,
Va.
ROSS, Mike. Supervisor of Operations at TMI-1. Placed in charge of operator activities in
the Unit 2 control room.
SCHEIMANN, Fred. Shift Foreman at TMI-2. Present in control room during early stages
of accident. Was in the auxiliary building when the accident began. One of four responsible for ini-
tial response to accident.
SCRANTON, William. Pennsylvania Lt. Governor and Chairman of the Pennsylvania Emer-
gency Management Council, which directs Pennsylvania Emergency Management Agency. Took
lead in State's emergency response on first day.
SEELINGER, James. Superintendent at TMI-1. Given responsibility for Met Ed's Emergency
Control Station in Unit 1 control room.
SMITH, George. XRC's Chief Health Physicist in Region I. Coordinated Region I response
with Eldon B runner.
SNIEZEK, James. Director of NRC's Fuel Facility and Materials Safety Inspection, Office
of Inspection and Enforcement. Responsible for assembling and assessing radiological information
received by Incident Response Action Coordination Team on the first day.
STELLO, Victor, Jr. Director of NRC's Division of Operating Reactors. Member of Incident
Response Action Coordination Team. First among NRC's top officials to diagnose uncovering of core
and existence of superheated steam. Now Director of Office of Inspection and Enforcement.
THORNBURGH, Richard. Governor of Pennsylvania. Responsible for determining whether
an evacuation was necessary.
WARREN, Ron. Met Ed Engineer. Notified NRC Region I of accident and manned phone link-
ing Region I and the site during morning hours.
WEAVER, Douglas. Instrumentation Foreman at TMI. Involved in taking incore thermo-
couple readings and installing a device to widen range of hotleg readings during morning.
WEISS, Bernard. NRC's IRACT Communications Officer. Incorrectly told White House Situ-
ation Room and Department of Health, Education and Welfare that there was never a problem
keeping core covered.
WILBER, Howard "Mike". NRC, Field Communicator at Incident Response Action Coordi-
nation Team.
WILKERSON, Scott. Nuclear Engineer at TMI. One of three engineers onsite during first
three hours of accident. Asked by George Kunder to analyze whether the reactor was going critical
again.
WRIGHT, Thomas. Instrumentation Technician at TMI-2. One of four technicians who took
incore thermocouple readings.
YEAGER, Bill. Instrumentation Technician at TMI-2. One of four technicians who took in-
core thermocouple readings. Based on readings, concluded core uncovered at time readings taken.
ZEWE, William. Shift Supervisor in charge of both TMI-1 and TMI-2. On duty when accident
began. One of four responsible for initial response.
91
The NRC Commissioners testify before the Subcommittee on Nuclear Regulation
Chapter 7
Accident At Three Mile Island:
The First Day
INTRODUCTION
At 36 seconds past 4 KK) a.m.. on March 28. 1979,
several valves in the secondary system of Unit 2
at Three Mile Island malfunctioned, causing first
the turbine and then the reactor to trip. 1 These
minor problems were compounded by yet another
valve that malfunctioned, this one in the primary
coolant loop of the plant. But it. too. was a minor
event. Safety systems came into play, as pro-
grammed, to control the situation.
Despite the correct functioning of the safety
systems, a variety of other factors complicated
the situation in such ways that the operators
were unable to respond effectively, and a serious
accident resulted.
It was a week before the plant could be declared
"stable." That week was characterized by further
problems, among them offsite releases of radiation,
a recommendation for protective evacuation, the
possibility of a hydrogen explosion and tremen-
dous anxiety among local residents. By the end
of the week, the Unit 2 facility was known to have
been severely damaged. How severely damaged
could not be determined because high levels of
radioactivity inside the containment precluded
entry.
The events of that week were largely deter-
mined by the damage done to the reactor in the
first two or three hours. During that initial period,
utility personnel had been unable to diagnose what
was happening and, therefore, took incorrect ac-
tions. What began as a routine incident very rap-
idly escalated into a major and serious accident,
although just how serious was not discovered un-
til two days later. The inappropriate decisions
and actions taken in the early hours were com-
pounded by the failure to diagnose plant condi-
tions further into the accident and by improper
actions throughout the day on the part of the
utility, the NRC and the State.
Because of the importance of what happened
during the first day and the need to insure proper
response during the critical, early hours of an ac-
cident, the Special Investigation focused on that
period. This chapter recounts and analyzes the
events of those hours and the responses of the
utility, the NRC and the State.
4:00:36 THE BEGINNING
Four men were on duty in Unit 2 at Three Mile
Island in the predawn hours of March 28, 1979:
William Zewe. Station Supervisor: Fred Schei-
mann. Shift Foreman for TMI-2: and Edward
Frederick and Craig Faust, control room opera-
tors. Each was a graduate of the Navy's nuclear
training program and had had at least five years
of Navy experience. All four had been through
Met Ed s training program, which included five
to nine weeks of practice on the Babcock & Wil-
cox simulator, and all had been licensed as plant
operators by the NRC. (1)
At 4 .-00 a.m.. Frederick and Faust were in the
control room performing routine duties. Zewe
was in the shift supervisor's office at the rear of
the control room. (2) Scheimann was in the tur-
bine building overseeing maintenance on the
plant's troublesome condensate polishing system.
1 For a description of plant equipment and plant systems, see "How the Plant Works," pp. 23-31.
93
As had happened in the past, a polisher had be-
come blocked by resin. Scheimann and his crew
were trying to break up the blockage with a mix-
ture of air and water. (3)
At 4 :00 :36 a.m., a year to the day and the hour
since TMI-2 had first gone critical, (4) valves
in the condensate polishing system malfunctioned
and shut off the flow of water to the feedwater
pumps. 2 The feedwater pumps, responding to the
lack of flow, automatically closed down, stopping
the flow of feedwater to the steam generators.
The pumps for the emergency feedwater sys-
tem, a back-up safety system designed for this
kind of equipment failure, started automatically
to pump water toward the steam generators.
However, closed valves in the feedwater lines
stopped the flow from reaching the steam
generators.
THE TURBINE TRIPS
With no water going to the steam generators,
insufficient steam was produced to run the turbine.
At two seconds into the accident, 4:00:38, the
plant's safety system automatically shut down
(tripped) the turbine in response to the feedwater
pump trip.
In the control room, Faust heard the alarms sig-
nal the shutdown of the main feedwater pumps
and said to Frederick, "Something's going wrong
in the plant." (5) Zewe came out of the shift
supervisor's office and noticed the turbine had
tripped.
With no water going into the secondary side of
the steam generators, not enough heat was being
removed from the primary system. The tempera-
ture of the coolant went up, and pressure in the
primary system began to rise as the rapidly heated
water expanded. Pressure in the pressurizer rose
to 2,255 pounds per square inch (psi), 3 100 psi
more than normal.
About three seconds after the start of the acci-
dent at 4:00:39 the pilot-operated relief valve
(PORV) atop the pressurizer opened auto-
matically to relieve the mounting pressure. Steam
shot out the valve and flowed into the reactor
coolant drain tank in the containment, where it
condensed into water.
THE REACTOR SCRAMS
Pressure inside the reactor vessel continued to
rise, triggering another automatic safety re-
sponse: at eight seconds into the accident
4:00:44 the control rods automatically dropped
down into the core, and the reactor "scrammed,"
terminating the fission reaction instantaneously. 4
As a result of the reactor scram, the heat being
generated by the core decreased sharply.*' This
decrease in the rate of heating, in combination
with the continued dissipation of some heat
through the secondary system, caused temperature
in the primary system to drop. As it did so, the
coolant contracted, thereby reducing pressure; it
would reach 1,100 psi within 20 minutes after the
accident began and then fluctuate between 1,000
and 1,100 psi for the next hour or so. (7)
All this occurred by 4 :00 :49 13 seconds into
the accident.
About 16 seconds into the accident, an operator
in the control room noticed instrumentation in-
dicating that the emergency feedwater pumps had
been automatically activated. (8) No one saw the
two lights indicating that the feedwater valves
were closed, blocking the flow from the pumps to
the steam generators. 6 (10)
THE PORV FAILS TO CLOSE
By about this time 16 seconds into the acci-
dent and about 12 seconds after the PORV had
opened pressure in the pressurizer had de-
creased to 2,205 psi, the point at which the valve
was supposed to close. The indicator light in the
control room went out, a signal that power to
the valve had gone off. The operators assumed
the valve had closed. 7 In fact, it had stuck in the
open position. (11)
A LOCA IN PROGRESS
The situation had become a multiple-failure
accident. 8 More important, the plant was now ex-
2 It was later determined that water had entered the air lines, a problem similar to that which triggered an incident
in 1977. See "Prior to the Accident," pp. 64-65, for further details.
3 References to psi throughout this section are to pounds per square inch gauge. It is equivalent to absolute pressure
less the atmospheric pressure of 14.7 psi.
' See "How the Plant Works," p. 30.
6 When fission stops, the heat produced by the core drops dramatically, initially to about six percent of the heat
produced when the reactor is operating at full power. (6) The residual heat, known as decay heat, decreases with time.
"An operator told Special Investigation staff that a maintenance tag obscured one light. (9) There has been no
explanation for the failure to notice the other.
7 See "Prior to the Accident," p. 86, for a discussion of the PORV position indicator.
8 When the condensate polishing system malfunctioned, it started what is called a loss of feedwater transient. This
was the initiating event. The closed valves in the feedwater line which blocked the flow of emergency feedwater to the
steam generators was the first failure in the unfolding event. The PORV sticking open was the second failure. The event
thus bcame a multiple-failure loss of feedwater accident.
94
Control room console showing maintenance tags
periencing a loss-of -cool ant accident, since the
failed PORV had become an undetected pathway
for coolant to escape the primary system.
Normally, equipment in the plant will auto-
matically detect the drop in a pressure that accom-
panies a loss of coolant and activate the Emergen-
cy Core Cooling System, which will control the
problem until it is resolved, if the system is left
to respond as designed. For a variety of reasons,
control room personnel did not diagnose the stuck-
open PORV and the resulting loss-of -coolant for
over two hours. In fact, they overrode the Emer-
gency Core Cooling System shortly after it came
on. A minor incident would soon become a major
accident.
At 41 seconds into the accident. 4 K)l :17. the op-
erators, as they had been trained to do when the
reactor scrams, manually started one of the three
make-up pumps that inject borated water into the
primary system in order to counteract the decrease
in pressure that typically follows a scram. 9 Boron
absorbs neutrons, further insuring shutdown of
the nuclear chain reaction. Zewe later explained
that this was a normal operator response to a feed-
water transient. (12)
In less than a minute after the reactor tripped,
the water level in the pressurizer had fallen from
its normal level of between 200 and 250 inches to a
low of 158 inches. (13) Pressure in the primary
system also continued to fall. This pattern was
typical of what happens after a reactor scrams,
and the transient seemed to be routine.
CONFLICTING SIGNALS APPEAR
At about this time, an unusual condition arose.
The level of the coolant in the pressurizer sud-
denly began to rise ; by six minutes into the acci-
dent', it would reach at least 400 inches. 10 (14) At
the same time, pressure in the primary system
continued to decrease. 11 The operators, not realiz-
ing the PORV was still open, would be confused
by these conflicting symptoms.
' One make-up pomp was running when the accident began. See p. 115 for a description of the relation between
uiake-up and high pressure injection systems.
** Actual levels could not be read since the scale only went to 400 inches.
11 Normally pressurizer level and primary system pressure move together.
95
Without the flow of feedwater, the secondary
side of the steam generators soon boiled dry, and
even less heat was being removed from the primary
system. The reactor coolant heated up still further,
expanding and pushing the water level in the pres-
surizer farther up. Pressure in the primary sys-
tem continued to drop.
HIGH PRESSURE INJECTION COMES ON
By two minutes into the accident, pressure in
the primary system had fallen to 1,640 psi, the
point at which the Emergency Core Cooling Sys-
tem is actuated. (15) The high pressure injection
system (HPI) , part of the Emergency Core Cool-
ing System, started automatically. 12 Two pumps
injected water from the borated water storage
tanks into the primary system at a combined rate
of 1,000 gallons a minute. This rate of flow was
fully adequate to compensate for the still-
undetected loss of coolant through the PORV.
Normally, automatic actuation of the HPI sys-
tem indicates a loss-of-coolant accident. However,
as described in the previous chapter, on several
other occasions, the HPI system at TMI-2 had
come on in response to less significant incidents,
and the operators had come to discount it as a clear
indication of a LOCA. 13
Further, the pressurizer level was continuing to
rise, a condition the operators found significant.
They had no direct way of measuring the water
level in the reactor vessel and therefore had to rely
on the water level in the pressurizer for an indirect
indication. Their training led them to interpret the
high pressurizer level to mean there was adequate
water in the primary system to cover the core.
(16)
A "SOLID" PRESSURIZER?
In fact, as the water level continued to rise
rapidly, the operators said they became worried
that the pressurizer was "going solid"- that is, fill-
ing completely with water. (17) This could cause
the steam bubble normally at the top of the pres-
surizer to collapse, which would in turn seriously
impede the operators' ability to control pressure
in the primary system. 14 Such a condition could
result in damage, possibly as severe as a rupture
in the primary system, (18) if there should be
sudden increases in pressure. Consequently, opera-
tors were taught to prevent the pressurizer from
going solid. (19)
THE OPERATORS OVERRIDE HPI
At 3 minutes and 13 seconds into the accident
4 :03 :49 the operators overrode the safety equip-
ment and took manual control of the HPI pumps.
At 4 minutes and 38 seconds into the accident,
4:05 :14, they greatly throttled the flow of HPI by
turning off one pump and cutting the other back
from 500 to about 25 gallons per minute. The opera-
tors also began to drain coolant out of the primary
system through the let-down line 15 at a rate in
excess of 160 gallons per minute. (20) By this
action, the operators also overrode the automatic
isolation of the let-down system. Water drained
through the let-down system is pumped into the
adjacent auxiliary building, which cannot be
sealed; this system later became a pathway for
radioactive releases.
Both actions throttling the HPI and draining
off the coolant were intended to lower the water
level in the pressurizer. By 4:06, the pressurizer
appeared to be solid. No matter what actions the
operators took, they could not reduce the water
level in the pressurizer significantly or reestablish
the steam bubble.
In fact, their actions were worsening the loss-
of-coolant. Water escaping through the stuck-open
PORV and the let-down system was not being ade-
quately replaced.
WHY HPI WAS THROTTLED
All the operators have stated that they throttled
HPI in response to the rapidly increasing pres-
surizer level, (21) that they had been worried the
system would "go solid." The level was reading
at least 400 inches (the top of the scale) and fluc-
tuated between there and 370 inches for most of
the next two hours. (22)
Scheimann, who by this time had returned to
the control room, recalled that when he gave the
order to throttle HPI, pressure in the primary
system was "low and stable." (23) Although
troubled by the low pressure, he said. "It would
have concerned me a heck of a lot more if it [the
pressure] was still going down." (24)
Michael Ross, the TMI-1 Supervisor of Opera-
tions who had come over to Unit 2 shortly after
the accident began, explained the operators' pre-
occupation with the pressurizer level to Station
Manager Gary Miller. At a review meeting held
by GPU two weeks after the accident, he said :
One thing on the pressurizer level that I
want to make sure you [Gary Miller]
" When HPI came on, it tripped one of the make-up pumps that was already operating, increased the flow of the
pump started earlier by the operators, and activated a third pump.
13 See "Prior to the Accident," p. 72.
14 The steam 'bubble acts as a cushion to dampen fluctuations in primary system pressure.
15 See "Technical Glossary," Appendix E, p. 371.
96
fullv understand. WeVe taught our oper-
ators, and we have a B&W written cau-
tion to never take the plant solid. Uncon-
sciously we have told them all the tune,
-never" take the plant solid." 1 (25)
A CONFUSING SITUATION
Even with HPI throttled and the let-down flow
increased, the control room personnel still could
not reestablish the steam bubble in the pressurizer.
Frederick later noted that the personnel realized
their attempt to lower the pressurizer level by
throttling the HPI flow "was not working./' (26)
We increased letdown, and we verified
the path from the bleed tank. "We thought
maybe our letdown passage was blocked :
that's why we filled up so fast. We tried
several things to try to establish pressur-
izer level. (27)
The problem with the pressurizer preoccupied
the control room personnel for much of the first
hour of the accident. In Scheimann's words :
... we sat there for quite a while with
pressurizer level up at the high end and
pressure holding constant at around 1100
to 1200 pounds. And it sort of, like sta-
bilized out right where it was at. Periodi-
cally. I could, by use of the letdown sys-
tem", get pressurizer level back down into
a visible range: however, it just wouldn't
seem to stay there. It would drift down a
little bit. then would go back up again.
(28)
Zewe could not understand why the pressurizer
level remained high despite the operators' efforts :
I didn't know where the water could be
coming from, except that if fmaybe] we
had some high-pressure injection valves
leaking: that were still feeding water,
even though we were throttling back I
did not know where the water was from.
(29)
He added that :
It was a real problem, in that I really
couldn't determine whv it was acting
that way. I really couldn't think of any
logical explanation. . . ." (30)
The control room personnel said that the re-
sponse of the primary system to the solid pres-
surizer also confused them. In Frederick's words,
. . . The pressurizer went full and we be-
lieved it [the reactor coolant system] was
full. It must have been full of water, but
the next confusing thing was the system
wasn't reacting as if it was solid. We
weren't seeing pressure spikes, so I dont
know if anyone concluded that there was
steam building someplace else. It was hap-
pening so fast, but we knew that we
weren't solid. 18 (31)
Frederick said further that at one point the con-
trol room personnel began to doubt the accuracy
of the pressurizer level gauge. (32) According to
Zewe, they checked several redundant level indi-
cators, requested a reading on the level from the
computer in the control room and had an auxiliary
operator check the level from a station in the aux-
iliary building. Zewe said they became convinced
the gauge was accurate. (33)
The control room personnel did not realize that
under certain conditions, pressurizer level cannot
be relied on to reflect the water level in the reactor
vessel.
Those conditions were present that morning at
the plant. They had also been present during a
previous incident in 1977, when steam became
trapped in the hotlegs, causing water level in the
pressurizer to rise, while pressure in the primary
system fell. The operators on duty at this time
were apparently unaware of this earlier incident. 19
Faced with two anomalous symptoms, one in-
dicating too much water in the primary system
(high pressurizer level), the other a condition in
which water was being lost or the primary system
was being cooled too rapidly (low primary system
pressure), the operators chose to respond to the
first. By throttling HPI. they had in effect con-
cluded that the problem was not the result of an
ongoing loss of coolant. 10
SATURATION IS REACHED
By about the time the pressurizer appeared to
be solid about six minutes into the accident
saturation had been reached in the primary sys-
tem : with pressure down, the water had begun to
boil. 11 The resulting steam bubbles in the coolant
14 The text of the caution appears in the addenda to this chapter. See Addendum 1, p. 153.
17 For further statements by control room personnel regarding the high pressurizer level, see Addendum 2. p. 153
* When the system is solid, pressure responds very rapidly to perturbations in the flow of coolant. As more water is
being pumped in, the instrumentation should show sharp increases in pressure, or spikes. These did not appear, confusing
the operators.
" See "Prior to the Accident," p. 65. for a description of this earlier incident which occurred during pre-operational
testing. Steam would become trapped in the hotlegs later on in the day.
" See Addendum 3. p. 153.
a When pressure is lowered, the boiling point of the coolant is lowered. As that point is reached, bubbles begin to
form in the water, also known as voids. This condition is called saturation.
97
displaced the water, pushing it into the pressurizer
and keeping the level up. The water being lost
through the stuck-open PORV was not being ade-
quately replenished, as the operators had throttled
the HPI pumps in an unsuccessful attempt to keep
the pressurizer from going solid. The core was on
its way to being uncovered. 22
FEEDWATER IS RESTORED
During this period, and for about the next two
hours, Faust was responsible for the feedwater
system on the secondary side, while Frederick and
Scheimann handled the primary system. (35)
Zewe supervised their efforts. (36)
Eight minutes into the accident, Faust realized
that no emergency feedwater was flowing into the
secondary side of the steam generators. He had
been checking the valves and discovered that a
pair of emergency feedwater valves "No. 12
valves" that were always supposed to be open
were closed. Faust opened them, thus reestablish-
ing the flow. (37)
It is generally accepted that the loss of emer-
gency feedwater for these eight minutes had no
significant effect on the outcome of the accident.
(38) However, it did add to the confusion and
distracted the operators as they sought to under-
stand what was happening. (39)
It is still not known when or why the valves
were closed. No TMI plant worker, operator or
supervisor has acknowledged closing them. Two
days prior to the accident, the system had been
tested, which required shutting and reopening the
valves. 23 The utility admits the possibility that the
valves were not reopened following this test. 24 (40)
Once Faust had established the flow of emer-
gency feedwater, he attempted to reestablish flow
in the main feedwater system to facilitate cool-
down of the plant. 25 However, the control room
personnel found other problems. 26 Zewe com-
mented that as the various difficulties became evi-
dent, "... I diverted a lot of my attention to
those items while they [Scheimann and Fred-
erick] were looking at the primary plant." (41)
CONDITIONS NOT UNDERSTOOD
At around 4 :20 a.m., Zewe left the control room
and went to the turbine building to try to fix some
condensate polishing equipment. (42) He did not
return until sometime between 4 :50 and 5 :00 a.m.
(43)
He described his general perception of the se-
verity of the accident at the time he left. His
statement shows that plant conditions were not
understood at this time :
. . . very soon into the accident and I'm
just saying within the first 5 minutes we
knew that we had an abnormal situation.
Then again there has not been a trip that
has really been textbook so to speak . . .
I didn't feel at this point in tune, that
the situation that we had, alright, was
. . . [a] very serious problem. But that
we did have an unusual situation with the
low pressure and a high level, ... I didn't
feel that we had anywhere near the scope
of seriousness of the accident that we
later developed into. . . . (44)
Faust did not realize what was happening
either :
The primary [system] at the time seemed
to have stabilized out with not a desirable
condition, but with I only remember as
being a high steam generator or a high
pressurizer level, and a low pressure, but
holding. (45)
Such perceptions and responses the operators
and managers incorrectly diagnosing the serious-
ness of the accident, people being absent from the
control room and attention focusing on relatively
less important systems or components while fail-
ing to recognize the significance of other condi-
tions would recur during the day.
It would be some time before anyone became
really concerned. For Zewe, it was not until
... we got into the point to where we
had to secure the cooling pumps or where
we chose to secure the reactor coolant
pumps. 27 (46)
22 Following the accident at Three Mile Island, the NRC issued a requirement that all operating reactors install
primary coolant saturation meters to provide readings on saturation conditions. Ironically, on February 26, 1980, Florida
Power Corporation's Crystal River-3 reactor was forced to shut down as a result of a loss of power related directly to
the installation of the instrument. The loss of power was apparently caused by a short in the electronics installed in
response to the post-TMI NRC requirement. The accident resulted in dose rates up to 60 R/hr in the containment building.
They declined to less than 0.2 R/hr in five hours. (34) See "Radiation Effects and Monitoring," p. 43, for a description of
the units of measure for radiation.
23 Closing the valves during testing while the plant is in operation was a violation of the Technical Specifications to
which the utility was legally obligated to adhere. The NRC has fined Met Ed $155,000 for this and other violations. See
"Recovery at Three Mile Island." pp. 210-211.
M On the basis of an FBI investigation, which found no evidence of sabotage, and because of limited staff resources,
the Special Investigation did not pursue the possibility of sabotage.
26 Cooldown involves removal of decay heat so that low temperature, low pressure conditions can be established in the
primary system.
26 See Addendum 4, p. 153, for a description.
" This occurred between 5 :15 and 5 :41 a.m. See pp. 104-105.
98
For Scheimann that recognition came at about
6 :30 a,m. :
. . . probably at the point where we were
starting to get the radiation monitors
and the different alarms . . . That was
the point where I was concerned that it
was more than an ordinary trip that we
had seen in the past. (47)
TOO MANY ALARMS
Within the first few minutes of the accident,
more than a hundred alarms had activated on the
overhead panels in the control room. (48) The
difficulties posed by this and other features of
the alarm system were familiar to the control
room personnel. 28
In Faust's opinion, the alarms "got in the way"
of the operators' efforts to diagnose the accident.
(49) Frederick said the operators ". . . disre-
gard[ed] generally the annunciator [alarm] sys-
tem as a whole, because it was not giving iis use-
ful information." (50) Zewe. when asked how
useful the alarm system had been in diagnosing
the accident, replied. "Xot very helpful." (51)
As noted in the previous chapter, the operators
had decided not to acknowledge the alarms acti-
vated during the initial stages of a complicated
transient until they had a chance to read them. 2 "
(52) However, in Zewe's words:
. . . the transient was so severe from the
standpoint of alarms, that [for] several
minutes, it was just intolerable to go
through each of the flashing alarms, so I
then acknowledged the alarms to silence
the alarm in the control room and just try
to handle the casualty based on plant
parameters, more so than alarm re-
sponses. (53)
Frederick said that after the first five minutes,
the alarms were activating at a much slower rate.
(54) Even at that slower rate, the control room
personnel differed on the usefulness of the system.
Zewe found that it was helpful:
Any new and subsequent alarms that
came in after that then were a lot more
meaningful because they came in at a
time fashion to where we could acknowl-
edge them and take action based on the
new incoming alarms. (55)
Frederick, on the other hand, said that it was
still "hard to tell" when new alarms were acti-
vated because so many were already lit and the
alarm noise does not change with additional an-
nunciators. (56)
M See "Prior to the Accident," pp. 67-70.
" See "Prior to the Accident," p. 69.
" See "Prior to the Accident," p. 71.
THE COMPUTER IS BACKLOGGED
The plant computer also proved of little use,
again as control room personnel had anticipated. 30
During the early stages of the accident, the com-
puter could not keep pace with the volume of
incoming alarms, and developed a one-and-one-
half hour backlog. (57) In addition, the paper
in the alarm printer jammed. (58) A post-accident
review revealed that none of the alarms activated
in the computer from 5:14 a.m. to 6:48 a.m. was
printed out. (59) Zewe hypothesized that the rec-
ord of those alarms, which should have been stored
in the computer's memory, were erased by a tech-
nician when trying to fix the printer. (60)
THE DRAIN TANK RUPTURES
Still unknown to the control room personnel,
steam and water were continuing to pour out of
the PORV and into the reactor coolant drain tank,
located in the containment. The tank was not de-
signed to collect flow for long periods of time,
since normally the valve opens for just a few
seconds.
About 4:04. pressure in the tank reached 150
psi, the point at which the tank's pressure relief
valve lifts. It did so, and steam and water, which
were at this point very slightly radioactive,
escaped into the containment. Pumps in the con-
tainment sump channeled the water into a waste
storage tank in the adjacent auxiliary building.
(61)
When the relief valve on the tank opened, pres-
sure in the tank leveled off for several minutes.
Then it began to rise again, (62) as water con-
tinued to pour in. When, 15 minutes into the ac-
cident, the pressure reached about 200 psi. an 18-
inch rupture disc at the top of the tank blew as
it was designed to. Pressure in the tank immedi-
ately dropped to just under 20 psi. (63) More
slightly radioactive water spilled onto the floor
of the containment. It, too, was pumped into the
tank in the auxiliary building. (64)
These very low-level radioactive releases were
the first from the containment.
As a result of the blown rupture disc and the
release of coolant into the containment, tempera-
ture and pressure in that building began to in-
crease. Pressure did not, at this time, reach the
point at which the containment automatically
seals itself, closing off the pathways, including the
sump pump lines, to the auxiliary building. Had
the Unit 2 containment been designed to seal auto-
matically upon actuation of the HPI. as it is at
some plants, the radioactive water would not have
99
been automatically pumped outside the contain-
ment. 31
Within 30 minutes, the tank in the auxiliary
building overflowed, releasing small amounts of
radioactivity into the building itself. Some of this,
in turn, escaped out the stack into the atmosphere.
At this time there was inadequate means of meas-
uring offsite releases. 32 It has since been calculated
that the releases were minimal and posed no health
hazard. 33
Response to Conditions in the Drain Tank
Many control room personnel were aware of the
increases in temperature and pressure in the con-
tainment and deduced that they had been caused
by the rupture of the tank. However, they failed
to recognize this as an indicator of an ongoing
loss of coolant through the PORV. (66)
Ken Bryan, a TMI-1 Shift Supervisor who ar-
rived in the control room about 4 :08 a,m., recalled
that shortly after he came into the room,
Somebody came around the corner and
said that the reactor coolant drain tank
was full and how about pumping it down ?
I walked around to pump it down and it
was empty. I looked and there wasn't any
water in it. The indication was downscale
all the way. I said oh ! oh ! and then I sort
of walked around front again. (67)
Bryan also recalled that at some point after
he noticed the loss of level in the tank, he heard
the containment fire alarm. (68) He said the op-
erators checked the temperature in the contain-
ment and found it was rising. (69) According to
Bryan, "... I think about this time we figured
that we blew the rupture disc on the drain tank." 34
(70)
Zewe also was aware that something had hap-
pened to the drain tank. He recalled that at ap-
proximately 20 to 25 minutes into the accident he
checked the gauges and noticed that the tempera-
ture in the drain tank was high, while pressure
and level were low. (71) Zewe surmised, "We
either had lifted the relief valve on it and it was
still open or we blew the rupture disc on it. Or
something else happened to the tank ... I didn't
know at that point." (72)
In another interview, Zewe said he also had
noticed that the pump which circulates water from
the tank through a cooling system ". . . had a very
low discharge pressure [which] means that we
ruptured the RC [reactor coolant] drain tank."
(73) A low discharge pressure means there is little
or no water in the tank. Frederick also was aware
of the low discharge pressure : "It didn't seem like
the pump was pumping." 35 (74)
George Kunder, the TMI-2 Superintendent for
Technical Support and the Duty Officer that day, 36
arrived in the control room at 4:50 a.m., having
been called about the turbine trip and reactor
scram. (75) He said that when he checked the con-
tainment pressure strip chart, it read "around 2,
2.2 pounds," (76) an indication ". . . that we did
have a pressure rise in the containment which
likely had come from the reactor coolant drain
tank rupture disc blowing. . . ." 37 (77)
On the other hand, others did not conclude that
the drain tank had ruptured. When Frederick
checked the instrumentation for the drain tank
about 40 minutes into the accident, pressure had
already returned to normal. (78) He said he did
not know that the rupture disc had already
blown, 38 (80) and he thought the monitoring in-
struments in the tank had been damaged :
I assumed that we just damaged all those
instruments by blowing the relief valves
in there. Okay. We either blew it dry or,
you know, a sudden surge of pressure
was too much for the instruments. Then
they failed. (81)
Temperature and pressure in the containment
continued to rise steadily after the rupture, going
from 120F to 170F and psi to 2.5 psi, respec-
tively.
The control room personnel failed to recognize
that the abnormal conditions in the drain tank,
and the resultant increases in temperature and
pressure in the containment, were caused by con-
tinuing loss of coolant through the stuck-open
PORV. Their statements indicate that many were
unaware of all the symptoms or of their sequence.
It should be noted that the instrumentation show-
ing the temperature, pressure and water level in
the reactor coolant drain tank was located on the
back of a panel in the control room, out of the line
of sight of the main console. Further, there was no
31 At TMI-2, the containment did not seal until 7 :56 a.m., in response to high pressure. (65)
32 See "Radiation Effects and Monitoring," p. 44.
" The releases were estimated through back-calculations that were supported by evidence developed by the Food and
Drug Administration. See "Radiation Effects and Monitoring," pp. 44-45.
34 See Addendum 5, p. 153, for other statements by Bryan.
M Ordinarily, the pump operates intermittently to remove water that collects in the tank from various leaks normally
occurring around pumps and valves in the primary system.
* He was licensed on Unit 1 and was studying to obtain his license for Unit 2.
37 For further statements on this matter, see Addendum 6, pp. 153-154.
"* In an interview conducted several weeks after the accident, Frederick said that "a few minutes" after the accident
began, he checked the instruments and noticed that pressure and temperature were high and the level was "down." (79)
If so. and had he believed the instruments, then he could have deduced that the rupture disc had blown when, 40
minutes into the accident, he saw that pressure in the tank was normal.
100
strip chart which recorded conditions in the tank
over time, making it difficult to reconstruct trends
and to connect the rupture of the drain tank with
the long-term loss of coolant through the PORV.
Without knowing the sequence, an operator aware
of one abnormal condition would not necessarily
see it in terms of a progression of events indicative
of an ongoing loss of coolant Further, as is evi-
dent from Frederick's statements, when the indi-
cators were checked subsequent to the rupture of
the disc, some conditions, such as pressure, were
back to normal. (82) Some personnel said they
were misled by this into thinking nothing was un-
usual. Frederick, for example, later explained that
since the pressure in the drain tank did not ap-
pear to be elevated, he did not suspect the PORV
was still open. (83)
In this same time frame, at about 24 minutes
into the accident. Zewe asked Bryan to get read-
ings of the temperatures in the discharge line lead-
ing from the pressurizer relief valves. (84) The
temperatures were high and were a further indi-
cation of flow into the drain tank. The actions and
statements of the control room personnel provide
no evidence that they understood the cause of what
they were seeing.
HIGH WATER LEVEL IN THE SUMP
The continuing loss of coolant led to another
symptom which appeared at this time, but. again,
its implications were not understood. Water had
been flowing into the auxiliary building for about
half an hour. At this point, an auxiliary operator
noticed that both sump pumps were running. The
auxiliary building storage tank was overflowing
onto the floor, and a control panel in the auxiliary
building showed that the water level in the sump
was high. (85) He reported these facts to the con-
trol room. (86)
Frederick got a sump level reading from the
computer: it showed 5.999 feet, the top of the
scale, suggesting that the actual level was off-
scale. 39 (87) When Zewe heard this it was about
40 minutes into the accident he ordered Fred-
erick to shut off the pumps. (88) Zewe said:
". . . and we knew at that point that . . . the
water from the RC drain tank was going into the
sump." 40 (89)
Persistent low pressure in the primary system
is another indication that a LOCA may be in
progress. Control room personnel had noticed this
symptom, but again they did not attribute it to an
ongoing loss of coolant. (90)
One reason for the confusion over primary sys-
tem pressure was that it had stabilized at a low
point about 30 minutes into the accident. Zewe,
for one, said :
I really did not feel that we had a loss of
pressure, anyway . . . [A]t this point in
time, we had a rather stable pressure con-
figuration, even though it was low. We
did not have a continuing loss of pres-
sure. (91)
Kunder, who had arrived at 4:50. said that he
found the situation confusing because he had
never seen pressurizer level pegged in the high
range with a concurrent low primary pressure. He
recalled that before these two parameters had al-
ways performed consistently. (92)
The operators were later to describe the accident
as a combination of events they had never experi-
enced, either in operating the plant or in training
on simulated emergencies. (93) All stated they
knew the combination of high pressurizer level
and low system pressure indicated an unusual
transient. (94) Zewe and Scheimann, however,
said they would have been more concerned had
pressure not stabilized, albeit at a low point. (95)
Zewe said that when he returned to the control
room just prior to 5 a.m., the operators began "try-
ing to put our heads together*' to come up with an
explanation for the accident. (96)
. . . All of us were talking together . . .
trying to come up with why the strange
indication. Everything looked very good
except pressure was low, and level was
high (97)
POSSIBLE ACCIDENT SCENARIOS
For about the next half hour the control room
personnel considered different scenarios, based on
symptoms described in the emergency procedures
that appeared to match the symptoms being ex-
hibited. 41 These symptoms included high pressur-
izer level, low primary system pressure, elevated
containment temperature and pressure, and an off-
scale high water level in the containment sump.
Radiation did not appear to be a key symptom
at this time. Control room personnel recalled only
one radiation alarm during this period at 5 :18.
(98) It was activated by an intermediate let -down
cooler radiation monitor that normally measures
radioactivity in the water in the secondary side of
the let-down heat exchanger. (99) Zewe did not
consider it significant. (100) He said he knew that
" This scale, like others, was calibrated for normal operating, not accident conditions.
** See also Addendum 7. p. 154.
" Emergency procedures are written instructions designed to assist the operator in responding to specific transients
and accidents. The procedure for a particular event lists the symptoms that that event is expected to produce. It also
specifies immediate actions and followup actions that the operator must take to respond effectively to the situation.
Statements by control room personnel on the use of the procedures appear in Addendum S, p. 154.
101
the monitors had very low setpoints, were very
sensitive and were located near the sump into
which the slightly radioactive water from the
drain tank had flowed. ( 101 )
The four possible scenarios (102) the control
room personnel remembered considering were :
A rupture in the steam line running from
the "B" steam generator ;
Leakage from the primary to secondary
system through the tubes 42 in this steam
generator ;
A break in the emergency feedwater line;
and
A LOCA. 43
CONSIDERATION OF A LOCA
With respect to a LOCA, the control room per-
sonnel had differing recollections aboiit whether
they explicitly considered it. Zewe said he never
did:
It really did not enter my mind that we
had a loss-of-coolant accident. I didn't
fully understand what I had, but I al-
ways think, in terms of a loss-of-coolant
accident . . . that your pressurizer level
is a big key. But I had the reverse of
what it would have been for a loss of
[coolant] level. (103)
He also misinterpreted another symptom:
I never really considered that we had a
LOCA. The automatic actuation of the
engineering safety feature system [high
pressure injection], I felt at the time
was because of feedwater initiation. (104)
Kunder likewise never considered that they
had a loss-of-coolant accident. (105)
However, Faust and Frederick said they postu-
lated a LOCA. Faust noted, ". . . we were looking
at possibilities of a LOCA for one thing." (106)
Others recalled referring to the LOCA Emer-
gency Procedure, (107) although it is not clear
exactly when. Frederick said they discussed the
LOCA Emergency Procedure in order to deter-
mine whether the accident involved a steam line
break or a loss of coolant. (108)
In the course of the accident, the operators re-
ferred to emergency procedures dealing with pos-
sible types of accidents. The procedure for LOCAs
appears to have been based on the assumption
that specific symptoms would become apparent
almost simultaneously at the beginning of the
accident. In effect, the procedure took a "cook-
book" approach which assumed that all the symp-
toms of a LOCA would be present unambiguous-
ly. It did not tell how operators should interpret
an ambiguous or different set of symptoms, nor
did it state which symptoms were the most im-
portant, to be responded to even if other symp-
toms were absent or seemingly contradictory.
However, it should be noted that procedures
are based on certain foreseeable circumstances and
are not meant to cover all possible situations or
to substitute for operator training and judgment
in unforeseen situations. In several other respects
the operators found the procedures to be vague,
confusing, incomplete and hard to understand. 44
A LOCA Is Rejected
The conclusion reached, according to Frederick,
was that the accident was not a LOCA. (109)
There were several reasons why operators failed
to interpret the symptoms as a LOCA. For one, in
their diagnosis of the situation they stressed the
importance of the seemingly unusual sequence of
three of the key symptoms of the accident : low pri-
mary system pressure, high containment pressure
and water in the containment sump. (110) They
expected that these symptoms would occur almost
simultaneously during a LOCA. (Ill) Zewe ex-
plained that their training for LOCAs led them
to look for these symptoms to occur "within sec-
onds of each other." (112)
In this case, there was a delay between when the
PORV was to have closed and when two of the
symptoms of the LOCA occurred. Because of the
size and location of the source of the LOCA the
stuck-open PORV the water and steam from the
leak remained in the reactor coolant drain tank for
15 minutes, at which point the drain tank rupture
disc burst, leading to high containment pressure
and water in the sump. By that time, the reactor
coolant system pressure had stabilized. Thus when
the operators became aware of symptoms such as
the high water level in the containment sump, they
did not relate them to the drop in primary system
pressure. (113) Instead, their statements"indicate
they viewed them as unrelated, possibly caused by
an event different from that which caused the ini-
tial drop in pressure. (114)
The Emerffencv Procedure states that one of the
symptoms of a LOCA is a "rapid continuing de-
crease" in reactor coolant system pressure. The
procedure does not explain, however, when or even
whether that decrease will level out. In a LOCA
involving a relatively small-sized break such as the
March 28 accident at TMI, the primary system
pressure would be expected to stabilize at a rela-
tively high level at some point after the accident
began, as it in fact did. (115) Not realizing this
" A break in a small primary pipe in the steam generator which releases radioactive primary water into the secondary
system.
" For further details on the choice of possibilities, see Addendum 9, pp. 154-155.
** See Addendum 10, pp. 155-156.
102
was typical, the control room personnel inter-
preted it to mean that a LOCA was not taking
place.
Second, although both Faust and Frederick
agreed that all the symptoms listed in a procedure
need not be present for the procedure to be consid-
ered applicable. ( 116) the absence of one key symp-
tom described in the LOCA Emergency Procedure
(117) led the operators to believe that the accident
involved something other than a loss of coolant.
The Absence of a Key Alarm
A key symptom referred to in the LOCA Emer-
gency Procedure it is described as a "unique"
symptom is the HP-R-227 radiation monitor
alarm. (118) It is critical, though not essential, for
diagnosing a LOCA. since the monitor measures
particulate and iodine gas radiation in the atmos-
phere of the containment. 4 *
Zewe. when asked what significance the HP-R-
2-27 alarm would have had during the first two
hours of the March 28 accident, replied :
[O]ne of the things you look for if you do
have a LOCA is that you have activity
indicated on the atmospheric monitor in
the building, so that certainly would have
been a key. (120)
Further, this symptom generally does not appear
in the event of either a steam line break or a tube
rupture.
Xone of the control room personnel said they re-
called the alarm from the HP-R-227 radiation
monitor coming on (it did not do so even around
6 :45 a.m.. when most of the other radiation moni-
tors were activated). (121) They said they had
specifically checked for it. According to Zewe.
I looked at the panel several times during
the first two hours into the accident and
the alarm would have been very evident.
Plus. I had a shift foreman [Scheimann]
that was right at the primary plant con-
trols which is directly across from the
alarm panel and I'm certain that he would
not have missed an alarm, because you
would have had to acknowledge it and it
has its own alarm sound. (122)
Frederick stated that according to the control
room personnels' interpretation of the emergency
procedure, the radiation alarm was the feature by
which LOCAs could be distinguished from two
other accidents having some characteristics similar
to a LOCA : a primary to secondary system tube
leak in the steam generator and a steam line break.
(123) Frederick said that when the operators re-
ferred to the "Symptoms" section of the LOCA
Emergency Procedure,
We had to decide whether what we had
was a large steam leak or a loss of coolant.
The difference between the two. according
to these procedures, is a radiation alarm
in the building, and a radiation alarm
never came in. (124)
Frederick said further :
We decided it was a non-radioactive leak,
therefore, it must be the steam system and
not the reactor coolant system. The symp-
toms are identical except for the radiation
alarm. (125)
Faust gave a similar explanation :
If you look at the diagnostic chart for de-
termining whether you have a steam leak,
or a primary leak, there is only one differ-
ence, and that is the radiation level. And
whether or not you're going to fall into
a LOCA procedure is determined by
whether or not you have a direct radia-
tion alarm. That's how the procedure
reads. There was no radiation alarm, we
were not in the LOCA procedure. That's
how it is. (126)
Problem* with the Monitor. According to two
XRC inspectors who analyzed the failure of the
alarm to sound, it may have been miscalibrated.
(127) Their post-accident review of the strip chart
for the monitor showed that from 5 :05 a,m. to 5 :25
a.m.. it was detecting levels of radiation above the
point at which the alarm was to activate. Larry
Jackson, an XRC inspector who studied the strip
chart, told the Special Investigation staff that it
is possible the level of activity detected by the mon-
itor did not go far enough past the alarm setpoint
to activate it in that period. (128) He also said
that if the alarm setpoint had been even slightly
miscalibrated, it would have prevented the alarm
from coming on. (129) He stressed that he had
not checked the monitor to determine if it was mis-
calibrated. 46 (130)
A second possible reason the alarm may not
have gone off was found later that morning, at
about 5 :50 a.m.. when the core was first becoming
uncovered. (131) Joseph B. Logan. Unit 2 Super-
intendent, and Richard W. Dubiel, Supervisor of
Radiation Protection and Chemistry, had joined
Kunder in the control room. Dubiel said that just
after he arrived. Kunder asked him to remove the
" The monitor is located in the auxiliary building ; (119) its readings appear on a gauge and strip chart in the control
room. If the readings go above a certain level, an alarm light on the control panel and a high-pitched horn will be
activated.
"The radiation monitor has been inaccessible because it is located in the auxiliary building, an area of high
radiation levels.
103
charcoal filter in the HP-R-227 monitor. (132)
Dubiel described what he found:
He fKunder] . . . was very interested in
getting a reactor building [containment]
atmosphere sample and in making prepa-
rations for a reactor building entry. I got
the technician, went down and we tried to
get a sample of HP-R-227, which is the
reactor building atmosphere monitor.
As we opened up the iodine monitor
holding a charcoal cartridge, a large
amount of water came out. I immediately
closed it back up, and with the amount of
water in there, my first thought was that
we had some type of a steam environment
in the reactor building atmosphere caus-
ing some condensation in the sample lines,
and getting water into the monitor. [S]o
I called George and told him we could not
get a sample off that monitor because it
was full of water. (133)
Larry Jackson, the NRC inspector, said that
water in the sample line would both have blocked
the air flow into the monitor and would have had
a shielding effect. Either would have reduced the
levels of radioactivity detected by the monitor.
(134)
A review of the monitor strip chart after the
accident provided support for this explanation.
(135) It showed that the amounts of radioactivity
detected by the HP-R-227 decreased until just
prior to 6 :00 a.m. (136) A decrease would not have
been likely, given conditions in the containment
during that time. A steady increase would have
been more probable. That decrease could have re-
sulted from the water in the monitor. (137)
The HP-R-227 monitors should perform well
in a steam environment, since they are relied upon
for diagnosing a LOCA, an accident which ex-
poses them to steam. This apparent design weak-
ness is of concern to the Subcommittee.
The Emergency Procedure Misinterpreted
The control room personnel focused on the HP-
R-227 radiation alarm monitor as the determi-
nant of a LOCA because the procedure charac-
terized it as a "unique" symptom of a LOCA
which could be used to distinguish between a
LOCA and a steam line break. The control room
personnel said they interpreted this to mean that,
other than for the alarm, the symptoms of a
LOCA and a steam line break were identical. (138)
The control room personnel were interpreting
the emergency procedure to mean that the absence
of this radiation alarm was conclusive evidence
that a LOCA was not taking place. This interpre-
tation is incorrect. Although a LOCA and a steam
line break inside the containment would produce
many of the same symptoms in the primary sys-
tem and the containment atmosphere, the second-
ary system would behave quite differently. For
example, pressure on the secondary side of the
steam generator would be expected to drop sub-
stantially during a steam line break, but not during
a LOCA. In short, there were symptoms other
than the HP-R-227 alarm that operators could
have used to distinguish the two accidents. 47
On the other hand, the Special Investigation
staff found that the wording of the procedure does
imply that an alarm on HP-R-227 is an extremely
significant symptom of a LOCA and that the
wording is broad enough so that the control room
personnel's confusion is understandable in that
context.
In another respect, the control room personnel
used the Emergency Procedure inappropriately.
Although the radiation alarm was not present,
neither was the "unique"' symptom of a break in
the main steam line. (139) If the procedures had
been followed strictly, both a LOCA and a steam
line break would have to have been precluded be-
cause of the absence of their unique indicators. 48
STILL MORE SYMPTOMS APPEAR
Around 5 :00 a.m., two more symptoms arose.
First, not long after 5 :00 a.m., the four reactor
coolant pumps began to vibrate excessively. As
noted, the coolant had become saturated, meaning
that steam bubbles had formed in it. To protect
the pumps from possible damage, plant operators
turned the first two off at 5 :14 a.m. 74 minutes
into the accident. (140) They did not take any
corrective actions to return the coolant to an un-
satu rated state, as they were not aware of that
condition. (141)
Then, at 5 :15 a.m., a reactor coolant sample was
analyzed to determine, among other things, the
concentration of boron in the water. The results
showed less boron than prior to shutdown, even
though the operators had been adding it contin-
uously since 4:00 a.m. (142) When the operators
and managers started getting higher neutron
counts from the source and intermediate range
monitors, 49 they interpreted them as confirmation
of the low concentration of boron. (143) They said
"This fact is reflected both in LOCA Emergency Procedure ("NOTE: 3," on pp. 6-7) and in the steam line
break Emergency Procedure. The unique symptoms of a steam line break are 1) low condensate storage tank level
alarm, and/or low hot well level alarm, and 2) Feedwater Latch System Actuation.
48 See also Addendum 11, p. 156, for another example of inappropriate use of the emergency procedures.
The nuclear instrumentation includes source and intermediate range monitors that measure the extent of neutron
activity (or flux) that is occurring in the core. The monitors are located outside the reactor vessel.
104
Top of the reactor showing instrumentation cables
they feared they \vere somehow diluting the boron,
thereby raising the possibility of recriticality.
(144) "
Shortly after the operators isolated the "B"
steam generator at 5 :27 a.m., 50 pressure in the con-
tainment began to decrease slowly. (145) The
change was coincidental, but led some control room
personnel to conclude that a steam line rupture
had in fact been causing the increase in pressure.
In Zewe's words, "So I said; I'll be darned, the
leaking generator was leaking into the building."
( 14-6 ) Scheimann commented, "I, myself, thought
that when we isolated the generator, we might
have stopped the leak." (147)
At about 5 :15 a.m.. Station Manager Gary
Miller called the Unit 2 control room and spoke
with Kunder. According to Kunder, he told Miller
he did not understand what was going on in the
plant. ( 148 ) As a result of this conversation, Mil-
ler directed that a conference call be established
between Miller. Kunder, Met Ed Vice President
for Generation Jack Herbein, and Lee Rogers.
B&W's site representative. 51
THE CORE IS UNCOVERED
At 5:41 a.m., still unaware that a LOCA was
in progress, the control room personnel took a
critical step. They shut down the last two re-
actor coolant pumps, which also were vibrating ex-
cessively. (149) This ended the forced flow of
cooling water through the core. So long as the
pumps had been running, the combination of water
and steam flowing through the core removed
enough heat to protect it even though coolant was
being lost. (150) Once the pumps were stopped,
the steam separated from the water and rose to
the top of the hotlegs the so-called "candy
canes" and the water level in the reactor vessel
dropped.
The uncovering of the core began very soon
after circulation stopped. Water was continuing
to escape out the PORV, at the same time that
HPI was being throttled, so that the lost coolant
was not being replaced. The water level continued
to decline, temperatures to increase, the coolant
to boil. Not only did the boiling release saturated
" See Addendum 9, pp. 154-15o, on isolation of the "B" generator.
" See p. 109.
105
5"-058 0-80-8
steam, which continued to rise toward the higher
portions of the system, such as the hotlegs, but the
steam displaced more coolant forcing it into the
pressurizer and out the PORV. This process would
continue to uncover the core.
The control room operators tried to remove the
heat by establishing natural circulation (i.e.,
convection flow) 52 through the primary system, a
method that would not require the reactor coolant
pumps. (151)
Neither Zewe, Faust, Frederick nor Scheimann
To Steam
Generator
Reactor vessel
' From Steam Generator
Figure A: Normal Conditions- Primary system
contains water
had ever used natural circulation to cool down the
TMI-2 reactor. (152) Although they had prac-
ticed initiating it at the Babcock & Wilcox simu-
lator, the practice did not continue long enough
for them to observe how a plant would respond
once natural circulation was established. 53 (153)
Their efforts proved unsuccessful because steam
had accumulated in the hotlegs, blocking the pas-
sage of any flow.
With saturated steam in the hotlegs and no
flow, a large difference in temperature developed
between water in the pipes going into the reactor
vessel the ''coldlegs" and water in the pipes
coming out the "hotlegs." Based on the evidence
reviewed by the. Special Investigation staff, the
control room personnel did not interpret this con-
dition to mean that there was no flow. 54
While the control room personnel were trying
to establish natural circulation, they lost a key
indicator needed to determine if it was taking
place. The hotleg temperatures went offscale: the
[|3 Primary Water
EFI Saturated Steam
o
To Steam
Generator
' From Steam Generator
Figure B: Primary system contains water
and saturated steam
Adapted from: Nuclear Safety Analysis Cer
temperatures in the "A" loop reached the high
point on the scale of (520 F at approximately 6 :10
a.m., the "B" loop at 6 :30 a.m. (154)
Frederick, trying to determine whether natural
circulation had been established and having lost
a key indicator, said that at that point :
. . . the only thing I figured I could do
was watch, hold the steam generator
levels up and watch the temperatures in
the steam generator and try and deter-
02 See "Technical Glossary," Appendix E, p. 372.
10 See Addendum 12, p. 156, for Zewe's comments on the usefulness of the emergency procedure in connection
with natural circulation.
54 In order to determine whether cooling of the core is taking place by natural circulation, the "coldleg" temperature
is subtracted from the "hotleg" temi>erature. The result is called "delta T," meaning the difference in temperature
between water in the hotleg and in the coldleg. If delta T falls within an appropriate range, it indicates that water
is flowing through the system, that the steam generators are not dry, and that they are removing heat.
106
mine a change in the delta T across the
core. Now we sat like that for I don't
know how long. ( 155)
WHAT WAS HAPPENING IN THE CORE
Hy piecing together the many analyses that
have l>een carried out since the accident, the actual
course of events during this period was recon-
structed. Normal reactor conditions are illustrated
in Figure A. Primary coolant fills the reactor ve>-
Primary Water
Saturated Steam
Superheated Steam
t
To Steam
Generator
Rom Steam Generator
Figure C: Al pumps off, reactor core drying out and
heating up. superheated steam flowing to hotlegs.
sel and flows smoothly through the system. Once
the coolant began to boil, however, saturated steam
wa- produced. '"'
Within minutes of shutting down the last two
reactor coolant pumps, at around 5:41, the top of
the reactor vessel was no longer filled with water,
but rather only with steam (see Figure H). As
the l(oilin<r continued, the water level in the vessel
dropped, progressively uncovering the core. Thus.
Ixiiling water surrounded the lower part of the
coiv. .-team the up|>er part. The exi>osed fuel above
the water level l>egau to heat up rapidly.
While steam will remove some heat, it is ineffi-
cient for this purl*).-*-. As the steam moved past
the ever hotter exposed fuel, it removed some of the
heat, becoming "sii|>erheated" in the process that
is, heated l>eyond the l>oiling point.
Water turns to steam and steam to superheated
steam at precise temperatures and pressures. The
American Society of Mechanical Engineers pub-
lishes standardized steam tables that show the
proi>erties of steam whether saturated or super-
heated, at varying teniix-ratures and pressures.
Steam tables were available in the control room at
Three. Mile Island on March -28.
Primary Water
Saturated Steam
Superheated Steam
and Hydrogen
To Steam
Generator
" From Steam Generator
Figure D: Core dryout and heatup continuing. Superheated
steam and hydrogen generated by zirconium/water
reaction collecting in hot legs.
The superheated steam rose to the higher parts
of the system, including the hotlegs (Figure C).
As a result, temi>eratures in the hotlegs rose
sharply. Once trap|>ed in the hotlegs, even if cool-
ant is injected into the core in sufficient quantity
to cover it again, the siqierheated steam may re-
main in the hotlegs. as it did at TMI."* The very
high hotleg temperatures that result are unique
signals that the core has been uncovered. (155)
Thus suj>erheated steam in the hotlegs can only
l>e interpreted to mean that the core has been un-
covered at some time, although it might not nec-
essarily lie uncovered at the moment.
Because the steam was not removing heat as
efficiently as water, the exj>osed fuel rods continued
tin Inilililes in a boiling ]*>t of water are "saturated'' steam.
* Snjierheated steam can be condensed into water by increasing pressure or lowering teiuieratnre.
107
to heat up. Within a short time, the Zircaloy clad-
ding around the rods reacted chemically with the
steam, severely damaging the cladding. The reac-
tions which were to reach their peak by about
6 :30 a.m. generated significant quantities of hy-
drogen, 57 some of which also collected in the higher
portions of the primary system, particularly in the
hotlegs (Figure D). This' hydrogen, together with
the superheated steam in the hotlegs, contributed
to the blockage of circulation through the system.
The remainder of the hydrogen escaped into the
containment through the PORV and the ruptured
drain tank. (158)
As the cladding around the fuel rods deteri-
orated, radioactive gases normally contained by
the cladding were released into the coolant. (159)
As the coolant flowed out the PORV, so did the
radiation.
If hotleg temperatures and primary system pres-
sure are known, steam tables can be used to deter-
mine if superheat conditions have been reached
in the system. Control room personnel knew by
6 :10 a.m. that the hotleg temperature in the "A"
loop had reached at least the upper limit on the
scale, 620F. At the same time, pressure in the
primary system was below 1,000 psi. At 620F,
any pressure below 1,780 psi is indicative of super-
heated steam conditions in the system. By using
the steam tables, the control room personnel had
the means available to deduce that there was super-
heated steam in the system. 58 Since it is not pos-
sible to have superheat without the core having
been uncovered, the control room personnel also
could have deduced that some portion of the upper
part of the core, had been uncovered. (160) At this
time, however, these conditions went unrecog-
nized. 59
As the morning progressed and the LOCA went
undetected, there was a continuing loss of coolant,
greater uncovering of the core, additional damage
to the fuel, and further release of radiation to the
coolant. (161)
RECRITICALITY A CONCERN
Because of the low boron concentration and high
neutron readings, the control room personnel said
they became concerned about increasing nuclear
activity in the core. (162) In addition, alarms
indicating low level radiation in the containment
were sounding periodically. The number of these
alarms increased considerably about 6 a.m.
At about that time, Michael Ross asked Scott
Wilkerson, 6 " a nuclear engineer who had been on
duty at Unit 1 at the time of the accident, to look
into the possibility of recriticality that is, the re-
sumption of the nuclear chain reaction, a develop-
ment that could have serious consequences. (163)
After Ross asked Wilkerson to look into the
possibility of recriticality, Kunder asked Wilker-
son to have Michael Benson, Unit 2's lead nuclear
engineer, was called and asked to come in to do a
post-trip review. 61 When Benson arrived about an
hour later, Wilkerson, another employee, and he
looked into the question of recriticality. (164)
STUCK-OPEN PORV IS RECOGNIZED
Around 6 :00 a.m., Brian Mehler, a Met Ed Shift
Supervisor, arrived in the control room. He noticed
that the pressurizer was full, or "solid," but that
system pressure had decreased, and concluded that
"at that point we had steam in the hotlegs." (165)
Mehler also noticed that the temperature in the
discharge line connecting the PORV and the re-
actor coolant drain tank was 229F, an abnor-
mally high reading. (166) This temperature was
a crucial indicator that the PORV was stuck open.
The PORV had been leaking since October 1978,
and the control room personnel had become accus-
tomed to abnormally high temperatures during
normal operations. Further, they knew the PORV
had lifted in the early stages of the accident.
Therefore, they did not conclude that the higher
temperatures were indicating a stuck-open
PORV. 62 (167) Although, in accordance with the
emergency procedure for PORV failure, the oper-
ators had requested temperatures for the PORV
and code safety valve discharge lines from the
control room computer twice before, they had dis-
counted the abnormally high temperatures. 63 (168)
The emergency procedure requires that the block
valve be closed if temperatures exceed 200 F.
(169) The utility personnel failed to do so, even
" At 1.600 F, the Zircaloy cladding of the fuel rods will react chemically with steam to produce hydrogen in an
oxidation process called a "zirc-water" or "zirconium-water reaction." (157)
M See p. 107.
" See pp. 117, 124-126.
" For the first three hours, Wilkerson was one of three engineers in the control room. This may have been significant,
since engineers have different training, are qualified to perform different types of work, and might have provided a
different perspective on the problem, all of which could have assisted the control room personnel in diagnosing the
situation.
81 A review of plant data following a reactor trip.
62 There is evidence that one operator used an incorrect method for diagnosing the failed PORV, using discharge
line temperatures. See Addendum 14. pp. 156-157.
63 A post-accident analysis revealed that the readings were requested at 4 :24 a.m., 24 minutes into the accident
(the PORV was 285 F, the code safety valves were 275 F and 263 F) ; at 5 :20 a.m., 1 hour and 20 minutes into it (the
PORV was 283 F, the code safety valves were 211 F and 218 F) ; and at 6:17 a.m., 2 hours, 17 minutes into it (the
PORV was 229 F) . See Addendum 13, p. 156, for further details.
108
though the readings they requested in the first
minutes of the accident were over that figure.' 4
At 6:22 a.m.. some two hours and twenty-two
minutes into the accident, and about a half hour
after he had arrived in the control room. Mehler
concluded the PORT was stuck open. 65 He recalled
that:
. . . what I saw was the pressurizer being
solid and no pressure in the system, pres-
sure going down. IT would indicate to me
at that particular time that either the
[pressurizer] heaters were not function-
ing or that we had a leak . . . And I asked
... if they checked if the heaters were on
. . . [and] I pushed out the temperature
for the electromatic [PORV] and codes
[safety valve?], and from that point. I
-umed that the electromatic was par-
tially opened because of the temper-
ature. . . . (174)
Mehler directed Scheimann to close the block
valve to isolate the stuck-open PORV.
Bryan's recollection of the overall situation in
the control room at that time was :
... we were kind of just . . . sit [ting]
back and started scratching [our] heads,
you know, trying to put this together.
That is about where we were at. (175)
Some control room personnel have made state-
ments indicating that at the time Mehler made
his diagnosis, they did not appreciate the reason-
ing for it. UTPii Scheimann later characterized
Mehler'* decision as "pretty much as [a] last
resort. . . ." (177) In Frederick's opinion :
As far as I know the action to close the
valve was . . . somewhat out of despera-
tion. In other words, there seemed to be no
other possible cause ... It [was] a last
ditch effort. (178)
Before the block valve was closed at 6 :22. some
> gallons of coolant more than a third of the
volume of the primary system had flowed out.
Damage already had been done to the fuel and
would continue to occur for at least another
hour. (179)
At the time the block valve was closed, the con-
trol room personnel took no immediate action to
replace the coolant that had been lost. (180) sug-
ting that they did not recognize that the plant
had experienced a loss-of-coolant accident.
Shortly after 6 a.m., Kunder at the plant and
three people offsite who would later play a major
role in responding to the accident Gary Miller,
the TMI Station Manager: Lee Rogers, the B&W
Site Manager: and John Herbein, Met Ed's Vice
President for Generation began a 35-minute con-
ference call. Herbein was the first utility corporate
executive contacted. He later contributed to the
successful eifort to stabilize the reactor late in the
afternoon. 66
The four men discussed conditions at the plant.
At one point in the conference call. Rogers asked
whether the block valve had been closed. Someone
was sent to find out and soon returned to say it was
shut. However, no one asked how recentlv it had
been closed. (181)
In fact, it had just been closed, a piece of in-
formation critical to anyone trying to determine
what conditions in the reactor vessel might be.
SUMMARY: FIRST 2Y 2 HOURS
The closing of the block valve brought the first
phase of the accident to an end. but there would
be further problems. The control room personnel,
in failing to diagnose the struck-open valve, had
responded to symptoms of the accident in ways
that aggravated the loss of coolant, resulting in
severe damage to the core. At this point, however,
none of them realized the core was uncovered.
The key stumbling block in the early attempts
to diagnose the accident emerged within the first
few minutes: the conflicting symptons of high
water level in the pressurizer and low pressure in
the primary system. The operators opted to ad-
dress the former, in effect rejecting the possibility
of a LOCA.
In part, this choice was a result of their train-
ing. Operators were taught to avoid collapsing the
steam bubble in the pressurizer. 67 Thus they
slowed the flow of the high pressure injection to
prevent the pressurizer from filling. They did so
at a time when, unknown to them, there was a
need to replenish the coolant being lost so that the
core would not become uncovered.
Their training had not adequately prepared the
cor.trol room personnel to deal with such problems
as multiple failures,* 8 plant behavior when
** The utility has been faulted for failing to follow correct procedures with respect to the prior PORV leakage.
TMI-2 Emergency Procedure #2202-1.5 ("Pressnrizer System Failure") contained sections on PORV and code safety
valve leakage. Orators were to respond to suspected PORV leakage by closing the block valve (170) and to suspected
code safety valve leakage by recording their discharge line temperatures on an analog trend recorder. (171) Prior to
March 2*. the utility neither used the recorder nor closed the block valve. It also did not repair the valve. (172) As noted
earlier, the XRC fined Met Ed .<l."o.OOO for these and other violations. (173) See "Recovery at Three Mile Island."
p. 210-211.
" See Addendum 15. p. 157.
" See p. 151.
57 See p. 96 and "Prior to the Accident." p. 74.
* See fn. 8, p. 94, and "Prior to the Accident." p. To.
109
natural circulation is used to cool the core, 69
and the absence of indicators of actual plant condi-
tions because of instruments going offscale. 70
Further, control room personnel did not con-
sider some of the symptoms to be typical of a
LOG A, again based on their training but also on
the emergency procedures, which they found to be
unclear, vague and incomplete. 71 Because they
neither heard nor observed a key indicator of a
LOCA, the HP-R-227 radiation monitor alarm,
which they mistakenly believed to be a necessary
indicator, they rejected the possibility of a LOCA.
Further, their training and the procedures led
them to believe all the symptoms would occur
within seconds of each other. During the accident,
the sequence was not as expected, and control
room personnel did not become aware of the symp-
toms as they occurred. They also failed to identify
the trends and relationships among the symptoms
that were indicative of a LOCA.
There were other problems : equipment malfunc-
tions; a lack of certain key indicators, such as
water level in the core ; poor layout of instruments
in the control room, particularly those relating to
the reactor coolant drain tank; too many alarms
coming on at once ; and a one-and-a-half hour back-
log on the computer. 72 Nor were the emergency
procedures helpful. They did not provide guid-
ance for decisionmaking in unforeseen circum-
stances.
Some of the problems can be traced to manage-
ment. The utility knew that one or more of the re-
lief valves on the pressurizer had been leaking for
six months. In such cases, the NEC requires that,
the utility either close the block valve or install
an analog trend recorder. 73 The utility took neither
step, nor did it identify or repair the leaking
valve.
Operator errors must be seen in the context of
these significant problems. Yet one person did di-
agnose the stuck-open POKV shortly after his
arrival in the control room. He did not, however,
initiate actions to replenish the lost coolant.
A SITE EMERGENCY IS DECLARED
During the conference call at 6:00 a.m., a
decision was made to try to restart the reactor
coolant pumps. Between this time and the at-
tempt to restart them at 6 :54, the severity of
the accident was to become clearer. For exam-
ple, about 6:30 a radiation technician began
surveying the auxiliary building. He found that
radioactivity was increasing rapidly. In the con-
trol room, radiation monitors for the contain-
ment and auxiliary buildings were showing the
same tiling. Alarms indicating high radiation
levels sounded in areas of the plant periodically.
(182)
Unknown to those at Unit 2, the core was
uncovered. Calculations made subsequent to the
accident show that temperatures in parts of it
may have reached 4,350-4,500 F, and possibly
higher. (183)
At approximately 6 :40 a.m., Dubiel phoned
Kunder, who was in the control room, to report
that two follow-up boron samples were showing
even lower boron concentrations 74 than the first
sample taken at 5:15 a.m. (184)
While Kunder was on the phone, radiation
alarms began coming in from all over the plant.
Kunder turned to Joseph Logan, Unit 2 Super-
intendent for Operations, and announced, in
very strong language, that they were "failing
fuel." 75 (185)
At 6:45 Zewe and Kunder declared a site
emergency, 70 as required by TMI's emergency
plan in the event of a possible "uncontrolled
release of radioactivity to the immediate en-
vironment."" (186)
NOTIFICATION OF OFFSITE AGENCIES
Ron Warren, a Met Ed engineer, arrived
in the control room shortly after the site emer-
gency was declared. Kunder directed that he
and Richard Bensel, another Met Ed engineer,
notify offsite agencies of the problems at the
plant, again in accordance with Met Ed's emer-
gency plan. Among those contacted were the
Dauphin County Civil Defense Agency, the
State Bureau of Radiological Protection and
the Nuclear Regulatory Commission Region I
office. (187)
" See p. 106.
70 See p. 106.
" See p. 102.
" See pp. 94, 96, 99-100.
73 A device that records temperatures over time.
74 See pp. 104-105.
75 Fuel failure means that the Zircaloy cladding of the fuel rods had been breached, allowing radioactive fission
products to enter the coolant. See p. 108.
76 See "Prior to the Accident," p. 79.
77 There probably had been no offsite release at this time. See p. 112.
110
According to Warren, he told those contacted
. . . that we had had a site emergency
and that we had possible fuel damage,
which is what George [Kunder] had
told me [and it] was about the only in-
formation he had related on to me, and
we thought we had a primary to sec-
ondary leak. (188)
Warren, when questioned by Special Investi-
gation staff as to exactly what he said, reit-
erated that he talked only of possible fuel
failure :
Question: Were those words used,
"The core may have been uncovered."
anything like that ?
WAPJIEX : Xo. those weren't. The only
words used were that we had possible
fuel [failure]. When we made the tele-
phone calls, we really didn't have that
much information. (189)
Warren stated that he had difficulty getting
additional information :
. . . Every time we tried to corner
George [Kunder] to get more informa-
tion, he was off somewhere else talking
to other people. (190) .
THE PUMPS WILL NOT RUN
At 6 :-"4 the control room personnel tried to re-
start (or "bump") the reactor coolant pumps. At
7:15 they gave up. Although the pumps started,
they would not run properly, since they were still
pumping mainly steam, rather than water. (191)
The inability to keep them going led to increased
recognition that there was steam in the primary
system. As Kunder told the Special Investigation
staff:
... I guess it was within mavbe the next
15 minutes, half an hour, when I. along
with everybody else, recognized that we
had significant steam void [ing] inside
the reactor coolant system. [The reactor
coolant pumps] did not produce any
flow ... it's apparent that it [the pumps]
was just spinning in a steam environ-
ment. (192)
During the attempt to restart the pumps, an-
other confusing set of indicators became apparent.
There was a sharp decrease in neutron activity in
the core and, at the same time, there appeared to be
a sharp decrease in the boron concentration in the
coolant. (193) Ordinarily, neutron activity would
increase as the amount of boron, a substance that
absorbs neutrons, decreases. 78
Kunder, recalling the drop in boron concentra-
tion from the 1,000 parts per million (ppm) meas-
ured before the accident to the 400 ppm at this
time, described his confusion during this period :
That really alarmed me . . . I was grasp-
ing at straws trying to assess what was
happening. So, initially, there I was feel-
ing we had a possible de-boration [ 79 ] of
the coolant system, and then we had the
400 ppm sample come in. I said. "Oh. my
goodness, it's still going." When we
bumped the reactor coolant pump, appar-
ently we let enough water into the core
[that the] intermediate range [neutron]
indications went down and source range
[neutron] indication went down, and I
said, "Ah ha. it's turned around." As
things evolved, it became apparent that
the indications were very confusing and
very misleading. 80 (195)
NEUTRON ACTIVITY
Around 7:15 Wilkerson turned to the issue of
recriticality that Kunder had raised with him just
before 6 KX) a.m. He and two newly arrived engi-
neers. Mike Benson and Howard Crawford, walked
around the control room checking the instrumenta-
tion panels and calling up information from the
computer. (196)
Benson described what he found to Special In-
vestigation staff. The hotleg temperature was off-
scale high, while the coldleg temperature was
abnormally low. (197) Pressure in the primary
system was down, and there was no flow because
the reactor coolant pumps had been turned off.
One set of neutron indicators outside the core sug-
gested normal levels of activity, but the computer
was providing high readings for another set in-
side the core. Normally, at the reduced power level
of the plant, the computer would not provide any
readings. 81
In an attempt to resolve these contradictory
readings of neutron activity. Benson checked a
backup set of neutron detectors that also took
readings from directly inside the core. The back-
up detector strip chart printed out data that also
showed high neutron activity. Benson concluded
71 The decrease in neutron activity was signalled by the neutron monitors. One explanation for the decrease is that
the core, which initially had been partially voided, was refilled to a certain extent when the pumps -were started. While
the core was partially voided, the neutrons had been able to escape the reactor vessel. When the core was refilled, they
were trapped, leading to a decreased signal. The converse may also be true : when the core was gradually becoming un-
covered, the neutron level rose proportionately as more neutrons escaped. (194)
Decrease in the concentration of boron in the coolant.
" See Addendum 16, p. 157, for Tx>gan's reaction to the behavior of the source and intermediate range monitor*.
n See Addendum 17, p. 157, for Benson's description of what was happening.
Ill
that the incore detectors had been made inoper-
able by excessive heat and that that had resulted
from a steam void in the core :
When I looked at the back-ups it indi-
cated to me how [the incore neutron de-
tectors] had slowed going back [down].
They had already gone through the void
and they had [seen] the worst case. They
couldn't recover. There is a temperature
limit [for the incore detectors] ... I
just assumed when the void went through
that it wiped them out. 82 (199)
OFFSITE RADIATION
Meanwhile, Crawford had, as required by Met
Ed's emergency plan, calculated a projected radi-
ation dose rate for Goldsboro, Pa., located directly
across the Susrniehanna Eiver. Using the proce-
dure prescribed in the plan for projecting doses,
Crawford made an extremely conservative pro-
jection, hypothesizing a high rate of radiation
leakage from the containment (0.2 percent of the
atmosphere in the containment per day) and ab-
normally high pressure in the containment (55
psi). (200) The rate came out at 10 rad per hour
(10 R/hr), 83 twice the level at which protective
action is mandated according to the EPA Man-
ual's Protective Action Guides. Had that been the
actual release rate, it would have necessitated pro-
tective action for Goldsboro and probably other
areas downwind of the plant.
At around this time, radiation monitoring teams
were sent to the site's perimeter and to Goldsboro
to monitor actual offsite dose rates, again in ac-
cordance with the utility's emergency plan. (201)
The releases, according to onsite measurements,
were small. P'urthermore, containment pressure
had not been greater than ? psi. (202)
EMERGENCY COMMAND TEAM SET UP
Gary Miller arrived in the control room around
7 a.m. As specified in the emergency plan, lie
assumed the role of emergency director and over
the next hour set up an emergency command team
to carry out TMI's emergency plan and to handle
the accident. (203)
Mike Ross, Unit 1 Supervisor of Operations,
was put in charge of plant operations, with Zewe
reporting to him. Dubiel was assigned the task of
radiation protection and monitoring. Logan was
to make sure that emergency plans were available
and being followed. Kunder was assigiied to super-
vise technical support and communications. Lee
Rogers. Babcock & Wilcox's Manager of Site Op-
erations, 84 who had also just arrived, was asked
to serve as the liaison with R&W and to provide
technical assistance. James Seelinger, Unit 1 Su-
perintendent, who would arrive later, was to head
the Emergency Control Station, which, after 10
a.m., was located in the TMI-1 control room. (204)
In addition to the members designated by Miller,
others such as Zewe and one or more of the NRC
inspectors who arrived later that morning partici-
pated in the team's meetings from time to time.
A GENERAL EMERGENCY IS DECLARED
While taking over as emergency director and
assembling the emergency command team. Miller
also focused on radiation monitoring. (205) Radi-
ation levels inside the, plant were continuing to
increase, and a potential for releases to the at-
mosphere existed.
At 7 :24, based on the radiation levels in the
containment measured by the containment dome
monitor. Miller declared a general emergency. 85
(200)
By approximately 7:30 a.m., control room per-
sonnel were increasing the amount of coolant being
supplied to the core. That action was producing
little additional flow. It was being inhibited by
the superheated steam and hydrogen gas trapped
in the hotlegs, which continued to prevent estab-
lishment of natural circulation. The trapped steam
and gas sustained the big temperature differentia!
between the hotlegs and the coldlegs. (207) More-
over, the reactor coolant pumps could not be
turned on because of the blocked pipes. Primary
system pressure, which stood at about 1.500 psi.
down from the 2,100 psi registered at around 7.
was being kept at that level by periodic venting
through the block valve into the containment.
(208) Damage to the core was already severe.
unknown to those at the plant.
82 The extent to which the incore detectors were knocked out (Indus' the early hours of the accident may not be
known until the core is removed. Another device the movable incore detector could have been used to determine the
operability of the fixed neutron detectors and as an indicator of the extent of uncovering. Xo one thought to nse it un-
til three days into the accident, partly because utility personnel considered it to be property of the reactor- vendor and
partly because of its status as "experimental.'' (10S)
" See "Radiation Effects and Monitoring." p. 44.
M Kabcock & Wilcox had provided the reactor. It is common for a reactor-vendor to assign a representative to a
plant using its reactor.
95 See "Prior to the Accident." p. 79.
112
INCORE TEMPERATURES
Sometime between 7:30 and 8 a.m., Miller de-
cided to use the incore thermocouples se for more
accurate temperature readings. (209) He said he
needed them in part to judge how effectively exist-
ing plant systems were removing heat from the
core, given that the operators had been unable to
establish natural circulation : (210)
. . . The context of what I was looking
for was a temperature indication that
would have some accuracy or be on a scale
of the instrument that I was reading ... I
was looking for a temperature on the hot
end to help evaluate neat removal from
an action standpoint. (211)
About 7:30, Miller asked his senior instrumen-
tation engineer. Ivan Porter, to get readings for
the incore thermocouples from the computer. (212)
The computer printed out nothing but question
marks. This meant either that the temperatures
in the core were greater than 700 F (the top of
the scale) or that the monitoring and readout
equipment was malfunctioning. (213) In fact, the
temperatures were greater than 700 F. There was
no other means in place for getting actual incore
temperature readings from the control room equip-
ment. (214)
The resistance temperature detector that meas-
ures hotleg temperatures also was registering off-
scale. It only told Miller that temperatures were
equal to or greater than 620 F. (215)
Eventually either Miller directed or Porter vol-
unteered to find another method of obtaining ac-
curate incore temperatures. Between 8 and 9 a.m.
Porter. Bill Yeager and Thomas Wright, two
instrumentation technicians, and Douglas Weaver.
an instrumentation foreman, went down to the
cable room. They were going to try to tap directly
into the wiring leading to the computer with a
device called a thermocouple reader. 87 (216) Por-
ter returned to the control room while the instru-
ment technicians tapped into the wiring. 88
Soon they were joined by another instrumenta-
tion technician, Bob Gilbert, and another instru-
mentation foreman, Skip Bennett. Weaver and
Wright left to install yet another measuring
device. Known as a "resistance bridge," it was to
be connected to the hotleg temperature detector
in order to extend the range of hotleg temperatures
that could be read from the control room. (222)
IS THE CORE UNCOVERED?
When Porter returned to the cable room, the
others had finished hooking up the thermocouple
reader. They got five initial readings (the device
could accommodate five thermocouples at a time).
These ranged from 200 to over 2,000 F. (223) .
When Yeager saw the 2,000 F reading, he said
he concluded the core was uncovered. (224) He
told NRC investigators that he made that state-
ment to those present. (225) Bennett concluded
that the core had been uncovered, but no longer
was. (226)
Wright and Gilbert, on the other hand, stated
that they did not think the core had ever been
uncovered. (227) Wright thought the thermo-
couples had been damaged. 89 (229) Gilbert, along
with Porter, believed the readings simply meant
the thermocouples were not functioning properly.
(230)
According to Porter, someone in the cable room
suggested they take additional thermocouple read-
ings by means of another instrument, a digital
voltmeter. (231) It provides a direct reading of
the voltage being produced by the thermocouples.
With the aid of a conversion chart, these readings
can be translated into temperatures. Porter, be-
lieving the thermocouples had been destroyed, told
the others he did not think it would be worthwhile
to use the voltmeter. (232)
The others did so anyway. Between 8 and 9 a.m.,
they took all 52 incore thermocouple readings with
the meter. (233)
THE INCORE READING IS DISCOUNTED
Just prior to 8 :15 a.m.. Porter had a brief con-
versation with Miller about the incore thermo-
couple readings. Miller told the Special Investi-
K Temperature measuring devices located in the reactor vessel a few inches above the core.
"' The incore thermocouples transmit their information to the computer through wiring in cables going from the
containment to the cable room, one floor below the control room.
M Recollections of who went down to the cable room with whom and when vary slightly. Wright recalled that
initially lie and Yeager went down to the cable room alone and were later joined by Porter and perhaps Bennett. He
recalled Gilbert having l>een involved originally in providing the thermocouple reader, but did not recall his presence
in the cable room. (217) Yeager also recalled going to the cable room to install the thermocouple reader with Wright,
having been directed to do so by Weaver. Subsequently, according to Yeager, Bennett. Gilbert and Porter arrived. (218)
Gilbert recalled going down with Bennett and finding Porter and two technicians taking readings. According to Gilbert,
by the time he and Bennett arrived, the readings taken off the thermocouple reader had already been acquired and the
digital voltmeter had been set up. (219) Weaver said he went down with Porter and had taken two or three readings
before Bennett and Gilbert arrived. (220) Xo one else recalled Weaver's presence in the cable room. Porter stated that
he went down with Bennett. Wright and Yeager. went back to the control room while the thermocouple reader was being
hooked up. then returned to the cable room and learned of the five readings. (221)
89 He said he thought the thermocouples had formed junctions with neighboring ones and that those reading over
2.000 were reading twice the actual temperatures in the core. He knew that temperatures of 1,000, while high, were not
high enough to indicate core uncovering and damage. (228)
113
gation staff that at the time, he was focusing on
Crawford's high projected offsite dose rate and
was waiting for a report from the offsite monitor-
ing team at Goldsboro. (234)
His conversation with Porter was brief. Accord-
ing to Miller, Porter gave Miller the five readings
taken off the thermocouple reader but said he did
not believe the thermocouples were reliable. (235)
Porter said that because he did not see any value
in the use of the digital voltmeter, he did not wait
for a full set of readings. 90 (239) Nor did Porter
tell Miller that two of his instrumentation staff
had concluded the core was then or had been un-
covered. However, Porter said he did not recall
having heard them make those statements. (240)
HOTLEG TEMPERATURES
Sometime between 8 and 9 a.m.. Porter, Weaver
and Wright finishing hooking up the resistance
bridge in the control room to the hotleg tempera-
ture detector. They then obtained actual hotleg
temperatures, which ranged between 680 and 720
in one hotleg and between 7(50 and 790 in the
other. (241) Although these temperatures indi-
cated superheated conditions in the hotlegs and
therefore uncovering of the core, the control room
personnel did not interpret them that way. (242)
It is unclear precisely when Miller and other
control room personnel received the hotleg tem-
peratures critical indicators of the condition of
the core. For example, the Special Investigation
found no record that established whether the
temperatures were available to Miller at 8:15
when, having established the utility's emergency
management structure, he assembled his key ad-
visors for the first of a number of "think tank"
meetings, a caucus approach to managing the ac-
cident that was characteristic of much of the first
day. Until interviewed by Special Investigation
staff on September 28, 1979, Miller said that he
was not even aware that the two devices had been
used to acquire thermocouple readings. (243)
THE EMERGENCY COMMAND TEAM
The meetings of the emergency command team
took place in the shift supervisor's office at the
rear of the control room. The first occurred at
8:15 a.m., four hours after the reactor had
tripped, three hours after Kunder informed
Miller that he was concerned that he did not
know what was going on in the plant, two and a
half hours after the core was first uncovered, and
nearly an hour and a half after the declaration
of a site emergency.
At this early caucus, the management team
established three general goals for handling the
emergency and bringing the plant to a safe and
stable condition: (244)
Protect the public
Keep the core covered
Protect Met Ed plant and personnel.
HPI: DEALING WITH UNCERTAINTY
In hindsight, the issue of greatest significance
at the meeting was high pressure injection, since
it was the only means of cooling the core. Earlier
in the day the operators had been confused by the
conflicting signals of high water level in the pres-
surizer and low primary system pressure. Re-
sponding to the former, they had turned off one
of the HPI pumps and throttled the second. 91
Early in the meeting of the emergency com-
mand team, someone in the group, without Miller's
knowledge, had decided to turn off the remaining
high pressure injection pump. As a result, foi
about five minutes there was no flow of coolant to
the core. (245) Subsequently in the meeting. Mil-
ler made a crucial decision: HPI should not be
turned off completely from that point forward. 92
Two weeks later Miller was to attribute his deci-
sion to uncertainty: '"Based on the instruments we
had we didn't know whether the core was covered.''
(24(5) However, in interviews he has made ambig-
uous statements about when he first realized the
possibility that the core was uncovered. 1 '"
Two weeks after the accident, control room per-
sonnel were not sure how the decision to stop HPI
came about and whether, in fact, HPI had been
completely turned off : 94
SEELIXGEE: There was a period.
though, after one of the caucuses, we
were in the middle of a caucus, and we
sent somebody out to secure the makeup
00 Bennett had transferred all 52 incore thermocouple readings to a computer sheet sometime between 9 a.m..
when the last measurement was taken, and 10 a.m.. when nonessential personnel were evacuated from the T'nit :>
control room. Wright said he saw the computer sheet in the instrument shop when he left the cable room. (236) Bennett
said he returned the conversion tables to the console in the control room, during whicli time be spoke to Porter and in-
formed him there were "several thermocouples that were extremely hot, in the neighborhood of 2.000 degrees.'' (237)
The computer sheet was not discovered until several weeks after the accident, when Bennett returned from vacation.
(238)
111 See pp. 96-08.
x For most of the day, HPI remained the most effective and, to a large extent, the only means of cooling the core.
M See pp. 124-129, for his other statements.
'* Plant data indicate that it was stopped at this time. (247)
114
pumps. 95 And we talked and they se-
cured the makeup pumps. We talked for
about two more minutes and Gary [Mil-
ler] came to the conclusion, we just de-
cided and I think it was through his im-
petus, that's the wrong thing to do. We
didn't totally understand it ...
ROGERS : Right. I do remember that.
SEELIXGER: . . . Let's go start the
makeup pumps again. That [sticks] in
my mind.
MILLER : In the room there I said [not
to] secure [expletive deleted] HPI.
ZEWE: We didn't secure the makeup
pumps we just secured HPI ...
* * *
a : Make sure that goes on the tape,
we never stopped the makeup pumps.
SEELIXGER : We never did stop the
makeup pumps?
Xever stopped the makeup
pomps.
SEELIXGER: Okay.
ROGERS : Xo. that's true.
SEKLIXGER: We sent somebody out of
the room with that intention and then we
changed our mind within a very short
period of time.
5 : Yeah, never did that, never did
that.
MILLER: Yeah. I was strongly in
(lisa . . . not in favor of stopping the
IIPIpun.ps. . . . (249)
Special Investigation staff later questioned Mil-
lei about his recollection of what had transpired at
:!." meeting.
Question : Can I ask who the individ-
ual was who was sent out of the room to do
it [sec-uiv HPI] and how the information
was jrott en to him or whomever not to do
it?
MILI.KI;: My memory is that the shift
su]x>r\ isor. Rill Zewe. and Mike Ro--.
weiv l>oth in the room when that direc-
tion was given. The man would have been
Rill Zewe who was in charge of the opera-
tion from the standpoint of the senior
watch supervisor.
Question: How was the information
gotten to him after he left the room that
he should not secure those pumps ?
MILLER : Mike Ross, who was in charge
of operations.
Question : And it was Ross who said
in the transcript. "Xever stopped the
makeup pumps." '.
MILLER : Yes,
Question : And then Seelinger says.
"Yes. we sent somebody out. but then we
changed our minds?" So that would be
what actually happened. Zewe went out
with the instruction to do it [secure
HPI], and then you changed your minds
and Ross was sent out to inform Zewe not
to do it. or to himself take the action nec-
essary to make sure that those pumps
were not secured.
* * *
. . . [M]aybe you can explain what
you do recall with respect to people com-
ing and going and what may have oc-
curred during that first caucus.
MILLER : During that caucus, the com-
mand group, as I have called it, were
making reports to me of activities in their
respective areas of responsibility. I be-
lieve during the caucus the shift super-
visor. Bill Zewe, came into the room and
talked to one of the members of the group
and not to me and then he exited the room
and subsequent to that I was informed
that high pressure injection was going to
be secured, and at that point I directed
Mike Ross to go inform and direct Bill
Zewe that high pressure injection pumps
were to be turned on and left on and
only turned off with my pei-sonal permis-
sion. (250)
WHO KNEW WHAT AND WHEN
Miller said his decision was based on the possi-
bility that the core was uncovered. (251) In the
same timeframe. others also had concluded that it
probably had been uncovered.
"The make-up pumi>s are actually the same as the HPI pumps, lu industry iirlance, make-ui> refers to a manually
rolled fl'iw rate, high pressure injection to the automatically delivered flow rate, which at TMI-2 was -~00 gallons
l>er minute per pump. "Securing HPI" means bypassing the automatic injection rate and operating the pumjis manually
at a lower (make-up) rate.
Tho mm ml room operators' actions make clear that they interpreted the direction "not to secure HIT' differently.
They neither increased the flow rate to the full-flow rate at which HPI comes on automatically, nor did they leave the
full-flow rate on when the high pressure injection system actuated at various points later in the accident. Prior to the
directive "not to secure HPI" having lieen issued, the operators had lieeu using one pump to provide coolant to the pri-
mary system. After the directive, the operators liegan using two pumps in a consistent fashion for the first time since
tlie accident began.
Their actions lead to the conclusion that they interpreted the directive "not to secure HPI" in terms of a distinction
lietween one pump ("make-up") and two pump ("HPI") operation. (248)
115
John Flint, 96 a Babcock & Wilcox engineer who
had arrived in the control room around 9 a.m.,
(252) and Bennett and Yeager, the two instrumen-
tation technicians who had been in the cable room,
indicated that they believed independently, and
with varying degrees of certainty, that the core
had been uncovered. (253) In fact, as noted be-
fore, Yeager said he believed the core was then
uncovered. 97 Flint told staff of this and other in-
vestigations that he told Lee Rogers, his manager,
of his conclusion shortly before 10 a.m. that
morning :
Question : Did you have any conversa-
tion about core damage with Lee Rogers ?
FLINT : Yes, I did.
Question : Was he in general agreement
with you ?
FLINT: We didn't discuss it in any
depth. I mentioned that we had core dam-
age, possible uncoverage of the core. At
that time, he was on his way to go into
a meeting in the shift supervisor's office.
We didn't discuss it further. (254)
Rogers was vague when questioned by the Spe-
cial Investigation staff on this point :
. . . [Flint] indicates that he mentioned to
me that he was sure we had uncovered the
core. And I did not recall that he ever
said that to me ; again, not thinking that
that was information that was going to
help me get the plant back to a stable
condition. I must reinforce that, because
he may very well have said it to me and
may have been very strong in his saying
it to me, but I did not recall that shortly
after [the accident], and I did not recall
quite a few months after. . . . (255)
According to Kunder, several others in the con-
trol room had surmised that the core had been
uncovered as early as 6:54 a.m., when the unsuc-
cessful attempt was made to restart the reactor
coolant pumps :
. . . We were concerned at that point that
we might be uncovering the core ... I
was concerned . . . that with the vapor
lock [the trapped steam] I just wasn't
sure in my own mind that all the flow was
going in through the core ... So I think
we were concerned for some indefinite
time, which may have been an hour or
two, that the core was indeed uncovered.
(256)
THE ROLE OF THE INCORE READINGS
Not everyone recognized the core was or had
been uncovered. One possible reason was the ex-
tent to which control room personnel doubted the
reliability of the incore thermocouples, which were
the only direct indicators of temperatures in the
core. At least four of Miller's six advisors Ross,
Logan, Kunder and Rogers were aware that
Porter had advised Miller that the incore thermo-
couples should not be considered reliable. Logan
said:
. . . He [Porter] had some [incore ther-
mocouple readings] that were high, some
low, and they didn't make any sense.
(257)
Kunder concurred :
. . . Based on the variation [in tempera-
ture values] he [Porter] didn't feel . . .
that the indications were reliable enough
to base any judgment or action on. (258)
Ross commented,
. . . [He said] not to take anything con-
crete off of them, that's what I deduced
from the conversation I heard. (259)
Rogers and Ross recalled actually overhearing
the discussion between Miller and Porter as it oc-
curred. (260) In an interview with Special In-
vestigation staff, Rogers described how the in-
formation was conveyed and how it was received :
. . . [The readings were conveyed] with
various given numbers relating to tem-
perature ; as high as 2400, as I recall, and
as low as a couple hundred degrees, with
a lot of them not reading, not giving any
indications at all, both with the individ-
ual readout and with the computer read-
out. And discussions, of course, being,
"Well, can we believe them ? Do we know
what they are telling us ? Are they really
good for this kind of an indication?"
That, again, when entered into some-
body's thought process once you inter-
ject into your thinking are you sure you
can believe them, is there any knowl-
edge or information you know that they
will perform in the kind of conditions
we have in the plant right now, you then
start not believing that any of them are
right. (261)
Miller said he had accepted Porter's opinion
that the incore thermocouple readings were un-
96 Flint had been assigned by B&W to the plant in connection with Unit 2 start-up operations.
97 See p. 113.
116
reliable. (262) Further, shortly after the 8:15
caucus with his senior advisors, he had gotten ac-
curate hotleg temperature readings, which met his
need for data to assess heat removal. He explained,
... I accepted [Porter's] advice and did
not go back and evaluate the specifics of
why he had said that [the incore ther-
mocouples were unreliable]. I accepted
his advice and at the same time I had an
indication of hot temperature that was
on scale on an instrument that I felt I
could depend on ... and a normally
used instrument, as opposed to the in-
core instrument, which is not recognized
or was not recognized at that time in any
[of our] procedures or training or test-
ing. So I didn't go back and question his
technical advice on that basis. (263)
Subsequent to the accident, Ross said he con-
cluded that the incore thermocouples probably
were the most reliable indicator at the time of
conditions in the core. (264) He explained why
he thought they were discounted :
. . . we have never trained our people [to
use them] nor do we use them, nor do
surveillance or readouts on those, saying,
"Hey. this is what you ought to be look-
ing at." We have iiever done that. I think
that's probably why it was easy to dis-
count them ... It was easy to discount
them, also because in many units [they]
are not even hooked up. In my unit
[Three Mile Island Unit 1] that's not
even connected, incore thermocouples.
(265)
THE MEANING OF WHAT IS KNOWN
Some control room personnel said they were
aware of conditions that clearly indicated the core
was uncovered, but that they did not make that
connection. For example. Miller, unlike Zewe and
Rogers, said he was relying on the hotleg tempera-
tures as an indirect measure of coolant tempera-
tures in the core. (266) The hotleg temperatures
that he said he had accepted as reliable were over
700 F. Given primary system pressure, which
Miller also had. those readings showed clearly the
presence of superheat conditions. (267) Miller's
subsequent actions in directing a futile attempt at
collapsing the steam bv repressurizing the plant
lead to the conclusion that he did not deduce that
superheated conditions existed or view the hotleg
temperatures as corroborative of the incore ther-
mocouple readings at that time. 98 Similarly, en-
gineers Benson and Crawford, who had been
" See pp. 124-125.
M See pp. 112-113.
focusing on the issue of recriticality, said they did
not equate significant steam voiding with uncover-
ing of the core. (268)
The evidence suggests another reason that some
control room personnel did not recognize the core
was uncovered : not everyone realized that there
was steam in the system or steam in the hotlegs.
According to Flint, as late as 9 a.m., Miller and
some of his advisors, may not have been convinced
there was steam in the hotlegs. Flint had discussed
this condition with some members of the manage-
ment team and did not believe at that time it was
generally accepted there was steam in the hotlegs :
Question: ... at that time, was the
belief that there were bubbles in the legs
shared by everybody else in addition to
yourself and Ed Frederick?
FLINT: No.
Question: And do you recall having
any conversations with any non-believers
regarding the existence of such bubbles in
the legs*
FLINT: Yes, I spoke with Lee Rogers,
Gary Miller, George Kunder, Bill Zewe.
Question : And all of those individuals
did not think that there were bubbles in
the legs . . . ?
FLINT: . . . [Prior to this] I did not
have the impression that they thought
there were steam bubbles in the legs. (269)
Kunder, however, said control room personnel
had deduced earlier, when they had failed to get
the reactor coolant pumps running, that they had
steam in the system. (270) Flint had also had a
discussion with Frederick shortly after arriving.
Based on the hotleg temperatures, the neutron de-
tector readings and other plant conditions, he said
the two had decided there was steam in the hot-
legs. (271)
RECRITICALITY NOT A PROBLEM
In this same general timeframe, Frederick and
Zewe spoke to Flint about the earlier concern
over jjossible recriticality. Flint said he concluded
that there had been voiding in the core and that
it had led to excessive leakage of neutrons from the
core which, in turn, had caused the sharp rise and
fall on the source and intermediate range neutron
monitors." According to Flint :
. . . [Ed Frederick, one of the control
room operators] mentioned that they had
earlier thought they had started to go
critical again. So did Bill Zewe [the con-
trol room shift supervisor] and one or two
117
other people. When I looked at it, I told
them that was not my opinion. I felt there
had been a change in the [neutron] leak-
age path from the core and that's what the
detectors were seeing. (272)
The source and intermediate range monitors and
other indicators were not signaling recriticality,
the main concern up to then, but simply a period
of time when there had been less shielding of the
monitors because of a steam void in the core. 100
Benson had recognized this :
. . . The problem had come up ...
"Well, do you think you have [gone]
critical again?" You know, they were
throwing around boron numbers like 700
ppm. I said, "That's ridiculous" ... I
assumed the void going through [the core
as] being the [cause] for the source range
being erratic. I also assumed the void had
messed up the incores . . . We shouldn't
have been critical at 700 ppm with all the
[control] rods in. (273)
According to Benson and Crawford, Flint was
present when the two of them were discussing the
excess leakage of neutrons from the core. In fact,
according to Crawford, it was Flint who led them
to conclude that the probable cause was voiding
in the core :
... It wasn't until ... we talked to John
Flint that we actually thought about void-
ing in the core causing the excess neutron
leakage . . . We were just kind of talking
and he said, "Well, that could be one of
the causes." We kind of agreed that that
would be a good cause. (274)
The Met Ed nuclear engineers Benson, Wilker-
son and Crawford not only concluded there was
voiding in the core, but also deduced the effect that
condition was having on hot and coldleg tempera-
tures. Benson said :
. . . The hotleg was really hot and the cold
[leg] was really cold, [and] that also
tended to make me feel that somewhere up
[in] the hotlegs . . . was the void. It had
already gone through the core and it was
somewhere up in the . . . hotlegs. Every-
thing seemed to look like that. I went over
[and] I talked it over with Scott [Wilk-
erson] awhile and Howard [Crawford].
We both tended to agree that that's what
had happened. (275)
Neither Benson nor Crawford concluded the core
had been uncovered. (276)
In summary, by 9 :30 a.m. certain control room
personnel had realized that the core had been un-
covered and that superheated steam had been pro-
duced. However, many others, including nuclear
engineers and the head of the emergency team,
still failed to recognize what was happening in
the plant. Thus at around 7 :45, when the utility
began communicating information on the accident
and on plant conditions to the NRC and State
officials, many did not know the extent of the
damage to the core.
NOTIFICATION OF THE NRC
When Miller announced the general emergency,
Warren and Bensel went through the same notifi-
cations as before, again according to Met Ed's
emergency plan.
NOTIFICATION OF REGION I
Met Ed first called the NRC Region I office at
7:10 a.m., but it was not until 7:45 a.m. that
Region I learned of the difficulties at TMI. When
Warren called, he had gotten only the answering
service, with whom he left a message. When the
Region I switchboard opened at 7 :45 a.m. and the
operator called in routinely, she got a message
from the service that a general emergency had
been declared at TMI, there was a primary to
secondary system leak in the "B" steam genera-
tor, and there had been an offsite release of radio-
activity. 101 The utility did not mention the
stuck-open PORV. (277)
The operator immediately called Eldon Brun-
ner, Region I Branch Chief, in his office. As he had
been told there had been an offsite release, he went
to the office of George Smith, the Region's chief
health physicist. (278) According to the Region
I Plan, in the event of a radiological incident,
Smith was the "appropriate branch chief" to
take primary responsibility for initiating the Re-
gion's radiological incident response program;
Brunner was the designee for operational inci-
dents. 102 (279)
In fact, the TMI accident involved both classes
of incidents. The plan did not specify who should
take overall responsibility in such a case. But this
did not lead to confusion, as Brunner and Smith
shared the responsibility. (280).
On his way to see Smith, Brunner stopped by
the office of the Regional Director, Boyce Grier,
and informed him of the general emergency. He
also directed that arriving personnel report to
Smith's office. (281)
100 When steam replaces water in the core, there is less shielding of neutrons, causing greater penetration of the reac-
tor vessel by the neutrons, which is picked up by the out-of-core neutron detectors.
101 In fact, there was no evidence of a release by that time. See "Radiation Effects and Monitoring," p. 44.
103 Incidents involving the operation of a reactor, in contrast to safeguard accidents or a release of radioactivity
without ongoing problems with the reactor itself.
118
From that office, both men returned the call
to the site and began recording information about
the accident. (282) They took the information
down on white notepads, rather than on the
"incident notification information" forms pre-
scribed by the Region's Plan. These forms speci-
fied what information was to be obtained: "the
cause of the incident,'' "the present status of the
material, facility or operation" and "actions taken
or proposed to be taken by the licensee." (283)
This kind of information was later not readily
ava liable to NRC headquarters.
While B runner and Smith were recording the
information, Donald Haverkamp and Richard
Keimig. respectively the Region I Project Inspec-
tor and Project Section Chief for TMI, arrived
in Smith's office. Their arrival permitted Brunner
to go back upstairs at approximately 8 a.m. to
activate the Regional Incident Response Center.
(284)
At some point the information Brunner had
obtained earlier was transferred from the note
pads to a blackboard, which served as the Center's
status board. From then on, all information was
put onto "Incident Messageforms," designed for
use in the Response Center. (285)
An open lino was set up with the TMI-2 con-
trol room at 8:10, and Warren began transmit-
ting information to the Region over it. (286)
NOTIFICATION OF NRC HEADQUARTERS
When Grier received the news from Brunner,
he called XRC headquarters in Washington, D.C.
and spoke with John Davis, Acting Director of
the Office of Inspection and Enforcement (I&E)
and a member of the NRC's Executive Manage-
ment Team (EMT). Grier advised Davis of the
general emergency and of the steps taken to set
up the regional Response Center. Davis then ac-
tivated the headquarters Incident Response Cen-
ter, comprised of IRACT and the EMT. (287)
At 8 :24. a direct line was established between
IRACT and Region I. The region transmitted
data from TMI to IRACT over this line.
One of the earliest pieces of information to go
| from Region I to XRC headquarters was the
temperature in the primary system. At 8:25 a.m.
Grier informed Xorman Moseley. IRACT Direc-
tor, that primary system temperature was 571 F.
(288)
This temperature was misleading, as it was
an average of both the cold and hotlegs ("T ve "),
rather than the significantly hotter temperature
of the hotlegs ("T h ") the primary system tem-
perature normally used to diagnose conditions in
the core. (The hotleg temperature in the "A"
loop was actually in the neighborhood of
680 F, 103 which, as noted, was indicative of super-
heated steam, given the pressure in the primary
system.) (289) Headquarters had no way of
knowing it was receiving an average reading.
NOTIFICATION OF THE COMMISSIONERS
Beginning at 8:37 a.m., Davis tried to notify
Chairman Joseph Hendrie and the other Com-
missioners. Hendrie was at a hospital in the Wash-
ington area where his daughter was having her
wisdom teeth extracted and for this reason was
out of the office all day. (290) He had infrequent
contact with the Commission and generally was
not directly involved in the agency's response dur-
ing the first day. 10 * (292)
Davis did not reach Commissioner Richard
Kennedy until 8 :53 a.m. at his office ; Kennedy said
that when Commissioner Victor Gilinsky arrived,
he would give him the news. Gilinsky was Act-
ing Chairman in Hendrie's absence. At 8 :57 a.m.,
Davis notified Commissioner John Ahearne, who
decided to go to the Response Center in Bethesda.
Shortly thereafter, Acting Chairman Gilinsky
called the EMT. 105 (294)
A Diagnosis of Core Uncovering
When Davis spoke with Commissioner Ahearne,
he told him that "a bubble was pulled into the
vessel." (295) Davis had given that same informa-
tion to Commissioner Kennedy a few- minutes ear-
lier. Ahearne asked that Edson Case, the repre-
sentative of the Office of Xuclear Reactor Regula-
tion on the EMT, be called to the phone to explain
what that meant. Case told him :
What it seems to signify to me is that
they lost enough coolant out of the pres-
surizer, and generally throughout the
system, that it apparently uncovered
part of the core and popped [the
cladding]. And I think probably the ac-
tivity so far is due to popping the fuel
elements. The real question is is that the
entire problem and have they regained
control over the primary system pressure
and level with their safety injection sys-
tem? (296)
Case's deduction and the question he put to
Ahearne were important. Not only did he diagnose
uncovering of the core; he also raised the next
step regaining control over the primary system
03 The offscale high reading was 620 F ; actual temperatures were 680" F.
101 Hendrie was aware of the accident as early as 10:05 a.m. when he spoke with Commissioner Gilinsky over the
phone about it. (291)
"The content of this conversation is unknown, as it was not present in its entirety in the IRACT/EMT tape
transcripts. (293)
119
by using the emergency water injection system.
(297)
There was no discussion at this time of the need
to consider evacuation or other protective action
in light of the possible consequences of an un-
covered core. 106 This is particularly relevant be-
cause, prior to Three Mile Island, it was believed
that prolonged uncovering of a nuclear core would
lead inevitably to a core meltdown. As Chairman
Hendrie testified in hearings before the Subcom-
mittee :
... I think I would have told you on the
27th of March that if you had a core sub-
stantially uncovered for some hours that
I would have to assume that major dam-
age, and probably some melting, was be-
ginning to go on ; and I couldn't tell you
with any confidence that it wouldn't con-
tinue to go on. (298)
THE FLOW OF INFORMATION
Warren continued to serve as the link between
the regional office and the plant :
... As long as I was up in the Unit 2
control room [until about 10 a.m. 107 ],
most of the morning was spent on the
phone with the NRC relaying messages
back and forth with them. . . . (299)
Assigned to the phone, Warren said he was never
fully briefed. He was unaware of important dis-
cussions that were occurring in the control room.
Because he himself only suspected as much, he did
not tell the NRC of the concern that the core was
uncovered :
Question : Were you aware on the
morning of the 28th that the core had
been uncovered at some point previously ?
WARREN : No, I suspected that it
had ... as soon as ... George [Kun-
der], my boss, related that we had pos-
sible fuel damage, I thought that there
was a possibility that we may have un-
covered the core. (300)
Core uncovering was not the only information
being mishandled. Misinformation about natural
circulation would be transmitted throughout the
day, starting at around 9 a.m. Even as Case was
posing the question about regaining control over
the primary system with the plant's safety
injection system, Kunder was answering it in a
simultaneous conversation he was having with the
regional office. Kunder reported to Region I that
there was a vapor lock in the hotlegs and that the
plant was not getting proper flow. Moreover, he
informed the NRC that utility management was
concerned the core was not being cooled. (301)
Within 10 minutes, this information was trans-
mitted to IRACT. (302) However, 10 minutes
later, the crux of Kunder's report that the util-
ity was having problems establishing natural cir-
culation was contradicted. Region I's George
Smith reported to IRACT that natural circula-
tion was being used to cool the primary system.
(303)
Then, less than 15 minutes later, Smith's report
was contradicted. Grier told Davis at the EMT
that the reactor was not being cooled through nat-
ural circulation and that the only mechanism for
cooling the core was high pressure injection. He
again informed headquarters that steam binding
was preventing natural circulation. (304)
Throughout the day, the transmittal of this sort
of contradictory information characterized com-
munications among the site, the Region and NRC
headquarters. Communications at times became so
confused that NRC headquarters was transmitting
contradictory information simultaneously. For
example, in one conversation at 1 p.m., IRACT
reported that there "seems to be all kinds of
bubbles in the thing; in one or two hotlegs and in
the core itself," (305) while at the same time, on
another phone, Victor Stello, NRR's IRACT rep-
resentative, was telling an aide to Congressman
Morris Udall that the reactor's primary system
was "water solid." (306)
There is evidence that the assumption that acci-
dents would be of short duration also contributed
to the communications problems. When the Re-
sponse Center was activated, IRACT was able to
open only one line of communications. (307) The
need for both radiological and operational infor-
mation quickly overburdened that single channel.
Around 2 p.m. on the first day an unidentified
speaker at IRACT would make the following ob-
servation :
This [accident] is interesting in the
broad sense that generally we always con-
sidered ... an event happening and
then a release. And this was a strange one
because we've got both going at the same
time. And it did create a problem at one
point . . . some people have said that the
reactor people were asking questions and
[the radiological] people were asking
questions, and they all had to wait in line
and there was some competition. Instead
of having a field [communicator] who
was [conversant in both], the field [com-
municator] they had . . . was good, but
he wasn't that conversant in health phys-
106 See pp. 132-135.
107 At that time, radiation levels in the control room forced non-essential personnel, including Warren, to evacuate the
Unit 2 control room for Unit 1.
120
ics stuff . . . there was a waiting in line
kind of thing [that] ended up [occur-
ring] , and he never thought about wheth-
er they were both primary. 108 (308)
NOTIFICATION OF THE STATE
At 7 :02 a.m. the utility had contacted Clarence
Deller, the duty officer at the Pennsylvania Emer-
gency Management Agency (PEMA), the desig-
nated lead agency of the State for emergency re-
sponse, to infornThim of the site emergency. Deller
in turn called William Dornsife, duty officer that
morning for the Bureau of Kadiological Protec-
tion (BRP), PEMA's technical arm, at his home.
(309)
Dornsife said he did not have the TMI phone
number and had some difficulty contacting the
Unit 2 control room. (310) Around 7:15 a.m. the
shift supervisor got a message from Dornsife and
returned his call. (311)
Dornsife was the only nuclear engineer in the
Pennsylvania Department of Environmental Re-
sources and therefore the only official in Pennsyl-
vania's State emergency organization who was
technically qualified to assess the status of the
reactor. (312) When the plant finally contacted
him, Dornsife went through a checklist of ques-
tions contained in the BRP emergency response
plan for the TMI site. It was designed to aid off-
site officials in assessing severity. (313) According
to Dornsife, he :
. . . asked other questions like status of
safeguards. Had the High Pressure In-
jection operated as designed ? Had the re-
actor tripped ? And they told me [yes] in
all cases. (314)
FLOW OF INFORMATION TO THE STATE
On the whole, Dornsife said he found it dif-
ficult to pin down what type of accident had oc-
curred because the utility was not sure what had
happened. (315) This difficulty also would reoccur
throughout the day.
The answers Dornsife received pointed to a
"Type 3" accident, as spelled out in the BRP site-
specific plan for TMI. (316) This type of accident
involved a release of radiation to the atmosphere
! as a result of system failures and included "design
i basis"' accidents, such as loss of coolant, and ab-
1 normal transients such as steam line failures or
i steam generator tube failures. (317)
By the time Dornsife was notified, the TMI
operators had already closed the block valve to
isolate the PORV, although they had not recog-
nized that the plant had experienced a loss of cool-
ant accident. Dornsife did not receive any infor-
mation indicating a "failure of the primary cool-
ant pressure boundary," 109 a symptom of a "Type
4" accident "major failure with failed safe-
guards." (318) Nor did the plant conditions he was
given suggest to him a failure of "engineered safe-
guards or mitigating features," 110 another Type 4
accident. (319) Accordingly, he said he did not
consider a "Type 4" accident. (320) At this point,
Dornsife said he saw no need to consider protective
actions seriously, nor did the utility recommend
that the State do so. (321)
NOTIFICATIONS BY THE STATE
In addition to notifying the BRP, PEMA also
notified others. A call was made at 7 :08 a.m. to the
Dauphin County Office of Emergency Prepared-
ness, 111 which had already been contacted by TMI
at 7 :02. At 7 :45, Col. Oran K. Henderson, the Di-
rector of PEMA, spoke with Governor Richard
Thornburgh, who at 8:10 called Paul Critchlow,
his Press Secretary, requesting that he start check-
ing into the incident. (322)
At 8 :20 a.m. PEMA notified Lt. Governor Wil-
liam Scranton. Scranton was Chairman of PEMA
and the person whom Governor Thornburgh put in
overall charge of State response to the emer-
gency. (323)
Throughout the early morning hours, Critchlow
and other aides in the Lt. Governor's office, along
with David Milne, Press Secretary of the Depart-
ment of Environmental Resources, and Dornsife,
began to assemble facts about the TMI crisis for
a previously scheduled news conference on energy.
(324) These individuals loosely formed what
amounted to an ad hoc emergency management
structure that, as the day progressed, had the effect
of minimizing PEMA's role in dealing with the
accident, as described below.
At 8:45 a.m. PEMA's operations officer, Dick
Lamison, notified the Federal Defense Civil Pre-
paredness Agency Region II of the accident at
TMI-2. That agency placed its health physicist on
alert and fed information to the other States
within Region II and to its national office. It
offered PEMA assistance, but Lamison said none
was needed. (325) PEMA also was in touch with
the Federal Protection Agency Regional Office in
Philadelphia. Again, it did not request Federal
assistance. (326)
As the day wore on, PEMA and BRP were to
call on outside resources, as did the NRC. Some
time after 11, Margaret Reilly, Director of
1 For two additional problems, see Addendum 18, pp. 157-158.
' A breach somewhere in the primary system, such as the stnck-open PORV, permitting a loss of coolant.
' These include safety features such as the Emergency Core Cooling System.
' This is another name for the county civil defense unit. Three Mile Island is located in Dauphin County.
121
51-058 0-80-9
BRP's Division of Environmental Radiation, ac-
cepted a second offer of assistance from Brook-
haven National Laboratory (BNL), which had a
radiological assistance plan under the aegis of the
Department of Energy. (She said she had rejected
their first offer of assistance at midmorning be-
cause she assumed the incident would be over be-
fore a BNL team could arrive.) (327) Brookhaven
sent a team to assist with radiation monitoring.
The BRP also relied on the National Weather
Service to trace and forecast wind speed and direc-
tion. The NRC Region I requested teams from
several other DOE field operations ; these were co-
ordinated through DOE's local command post,
established at Capitol City Airport, New Cumber-
land, 10 miles northwest of the plant. (328)
PEMA'S ROLE
From the beginning, PEMA as an organization
played a relatively minor role in the accident, de-
spite its designation as the lead agency and over-
all coordinator of the State's emergency response.
(329) There were several reasons. For one, PEMA
had no technical experts. After notification of the
accident by TMI, as outlined in PEMA's emer-
gency plans, PEMA no longer spoke directly with
the utility. (330) Instead, PEMA personnel were
to rely on BRP to provide and interpret data from
the site and to recommend the need for protective
action such as evacuation. (331)
However, the link between BRP and PEMA
proved very weak. PEMA logs reveal that PEMA
operations personnel received no new information
from BRP between 9:40 a.m. and 12:30 p.m. on
March 28. (332) According to Dornsife, BRP was
generally so busy with radiation monitoring that
staff often forgot to brief PEMA. He acknowl-
edged that BRP was supposed to keep PEMA
informed
. . . and get back to them and tell them
what the situation was to begin with . . .
[but] we were so involved in tracking,
we forgot to inform PEMA. (333)
Reilly agreed that, amidst its monitoring ac-
tivities, BRP may not have kept PEMA suffi-
ciently informed. She said she surmised, in addi-
tion, that BRP's data may not have been under-
stood by PEMA personnel because it was too
technical. She said :
I think we probably, to some extent, [fell]
down on the job with them in that we
didn't tell them information as to what
was going on. Our stance was that if we
perceived something they needed, we
would tell them . . . you try to tell them in
the language we speak, and it [the infor-
mation] doesn't go through the conduits
into PEMA too well. (334)
Conversely, PEMA's Deputy Director, Craig
Williamson, recalled that much of the informa-
tion PEMA received from BRP was general in
nature and unrelated to a determination of the
need for protective action :
We experienced difficulty [Wednesday
and Thursday] with getting information
in the form that we would disseminate it
to the emergency system in the affected
area. Much of the information was very
general in nature and lacked the specifics
that we really needed to inform the field.
(335)
In addition, PEMA was understaffed. Accord-
ing to John Comey, the PEMA Public Informa-
tion Officer, in an extended emergency, he, for one,
needed "trained personnel" who were aware of the
requirements of the press and in a position to
assist him. (336) In the past, the Governor's Press
Secretary had assigned him public information
counterparts from other State agencies. This time,
that assistance was not forthcoming. (337) He ex-
plained the situation :
... it was about two and a half months
into the administration . . . Members of
the Commonwealth press office, the
Governor's press office, were not aware
of what our function was as far as the
public information piece of it is con-
cerned. They were not aware of our capa-
bilities, the limitations, and the need to
coordinate, the fact that we had a history,
a good history, of providing this type
of service to the members of the working
press. (338)
A major problem for PEMA, however, was
relations with the Governor and Lt. Governor's
offices. When Comey and his temporary assist-
ants wanted to tell the press that PEMA was in
an "advanced state of readiness," that approach
conflicted with the Governor's. According to
Comey :
This type of descriptive adjective [ad-
vanced state of readiness] the Governor
did not want to use. He wanted it very
low-keyed . . . [T]o give the press the
confidence that we could accomplish this,
which I knew we could, the [our] de-
scriptions were a little more aggressive
and this was quite contrary to what the
Governor had in mind. And for that rea-
son, he was critical in the very early stages
of what we were doing down here. (339)
122
Critchlow restrained PEMA's public informa-
tion function :
Question : What was the thinking be-
hind what appeared to be a rather
hampered ability on [Comey's] part to
deal with the press?
CRITCHLOW : Early on ... Comey was
overloaded some of the people he was
drawing in to help were making some
potentially panic type inciting state-
ments so we moved [in] to make sure
any statements they did make were
cleared through me and my office. (340)
Comey commented :
. . . My hands were tied. I was not per-
mitted to conduct press conferences . . . to
conduct daily press briefings. So it was
[on] a one to one basis and that one to
one often would include as many as one
hundred to one hundred and fifty mem-
bers of the National and International
Press he re physically in the office. (341)
There was also some question within PEMA at
the beginning of the accident as to the accident's
severity. Upon notification, PEMA Director Hen-
derson's first reaction was that it was a test. (342)
After the Lt. Governor's first briefing, Hender-
son said he came away feeling the accident was
small, isolated and insignificant. (343)
The extent to which PEMA was excluded as an
active participant is evidenced by the fact that
it did not activate its Emergency Operations Cen-
ter until Friday. Until then, according to PEMA]s
Operations Officer, PEMA had received no indi-
cation that the problem at TMI was serious. (344)
Cxi VPII this situation, it was impossible for
PEMA to work effectively with local agencies.
Kevin J. Molloy, the Dauphin County Emergency
Preparedness Director, said the information from
PEMA was either so general, it was useless, or so
technical, it told the lay county civil defense offi-
cial nothing. (.>45)
Comey pointed out. however, that he did not
believe PEMA was supposed to transmit details
on TMI to the counties, even had PEMA had that
information. Nor did he believe the information
was needed :
The information provided to the Coun-
ties throughout this period was the type
of information they needed by and large
to perfect their responsibilities in the
evacuation process . . . Where we could,
mention was made of what conditions
were at the facility . . . The only thing
that was required was that they know the
task to be placed upon them, and also
know the time frame that we are talking
about. (346)
AD HOC STATE MANAGEMENT
The Lt. Governor's ad hoc management struc-
ture took over PEMA's role. (347) This group
was located in the State Capitol building and drew
on resources from State agencies, including BRP
and PEMA. The briefings and news conferences
often involved PEMA Director Henderson, BRP
Director Thomas Gerusky and Dornsife.
There was criticism of the failure or inability
of the ad hoc emergency group to transmit infor-
mation to others. Comey said, "The information
coming from the Governor's office was almost
non-existent." (348) Molloy put it even more
strongly. First he described the actual chain of
command :
. . . The accepted chain of command is
local-to-County-to-State-to-Federal . . .
When this procedure is followed, emer-
gencies are handled expeditiously and
professionally . . . Basically, through
the entire incident the chain of command
information-wise, went something like
this : TMI/NRC/Governors Office, at the
top block, the next block was the news
media and the public, and last but not
least, the Pennsylvania Emergency Man-
agement Agency and County and local
emergency personnel. (349)
Molloy commented that often he and others got
their news from the radio and television and that
lie was very dissatisfied with the flow of commu-
nications :
. . . We did not have time to listen to the
radio and T.V. etc. Yet this was the way
the information was coming out of the
Governor's office. It was not being given
to PEMA, who could have filtered it down
to us, etc. So information-wise from the
Governor's office. I feel it left a lot to be
desired. (350)
Molloy and others complained often, on one
occasion directly to the Lt. Governor, when he
visited Mollov's office in the Dauphin County
Courthouse. Although Scranton promised the flow
of information would be improved, Molloy did
not recall that happening. (351) He said he
blamed the Governor and Lt. Governor's offices
for undermining both the flow of information and
the predesignated chain of decisionmaking. (352)
123
DECISION TO REPRESSURIZE
As noted, between 9 a.m. and 10 a.m., only a
few people in the control room at TMI-2 had
realized the core had been uncovered. 112 General
agreement had finally been reached that steam was
present in the primary system, though there was
less agreement as to its location within the system
and whether it was superheated. Some believed
the steam was only in the hotlegs, others that it was
in the core as well.
Mehler, as noted, had concluded there was steam
in the hotlegs at about 6 a.m. Wider recognition
that there was also steam in the core had come
with the onslaught of radiation alarms at 6:40
a.m. and the unsuccessful attempt between 6:54
and 7:15 a.m. to get the reactor coolant pumps to
run. (353) Yet Miller, who was in charge of emer-
gency operations, replied that he was only aware
of steam in the hotlegs :
. . . By 9 a.m. I was convinced that we
had steam phase in the hotlegs because
of the [hotleg] . . . temperatures plus
the start of the reactor coolant pumps
which showed us they were not pumping
water. So I would say we were aware of
a steam condition. . . . (354)
There is no evidence that he deduced the core
had been uncovered.
On the other hand, John Flint said he recog-
nized that the steam in the reactor was so hot
(that is, so superheated) that it could not be col-
lapsed back into the coolant. 113 He also said he de-
duced that the core probably had been uncovered.
(355) At the time it was generally known that
the only way superheated steam could be pro-
duced in a reactor was through core uncovering. 114
(356)
Between 9 and 9:30 a.m. the Emergency Com-
mand Team decided to repressurize the reactor
in an effort to collapse the steam 115 and estab-
lish natural circulation. Flint said he argued
against this strategy, recognizing that the super-
heated steam was so hot in excess of 700 F,
according to the hotleg temperature readings
that pressure could not be raised high enough
within the capability of the system to collapse the
steam back into water :
. . . That morning I had recommended
against the repressurization because, if
the temperatures were true, we could not,
in fact, collapse the [steam] bubble. (357)
Nevertheless, the Emergency Command Team
directed the operators to repressurize.
REPRESSURIZATION FAILS
To raise pressure, the amount of makeup was
increased and the block valve kept closed. By
9 :45 a.m., pressure was about the same as at the
start of the accident 2,100 psi. However, the
steam bubbles did not disappear.
Flint had been right with temperatures at
700 F or greater, pressure would have to have
been raised around 3,000 psi to collapse the super-
heated steam, a pressure that exceeded the NRG
Technical Specifications and approached the
maximum test pressure for the system. 116 (358)
Superheat Went Unrecognized
The fact that the Emergency Command Team
attempted repressurization shows that its mem-
bers had not recognized there was superheated
steam in the system. There is no evidence to sug-
gest that superheated steam was ever discussed
at any of the emergency command team meetings.
Two weeks after the accident, when the man-
agement team gathered to record its recollections.
Rogers, Miller and Ross discussed the decision to
repressurize and the lack of insight into super-
heated conditions. The following exchanges clear-
ly indicate that this condition had not been a
consideration.
ROGERS : . . . Somewhere around 9 :30
you [the utility] started raising pressure
trying to collapse the steam bubble. In
our stupidity, we thought we could col-
lapse a superheated steam bubble and we
weren't even thinking it was superheated
at the time.
Ross : No, we were just trying to pour
water into the [legs].
ROGERS: We weren't even thinking
about it. We were just trying to push the
pressure up with [high pressure injection
water] being injected to try and get the
thing solid [with water]. We knew we
had steam in the loops, and we knew we
had to get it moved somehow. And that
was our attempt, after a meeting in the
supervisor's office, ... to raise pressure
la See pp. 113-114, 116.
115 See fn. 115 below.
14 The thesis that the only proximate cause of superheated conditions is core uncovering is now being challenged as
a result of events at TMI. Some analysts contend that, following uncovering of and damage to the core early in the
morning, superheat was produced in the afternoon by fission products in the primary system hotlegs. See pp. 142-143 on
conditions in the afternoon.
115 One way to rid the system of steam is to subject it to enough pressure that it is forced back into solution in the
water. This can be done by adding water to the system, which will cause pressure to increase, or by closing the system,
allowing pressure to build as heat to the system is increased.
16 The pressure at which the reactor coolant system was designed to operate was 2,500 psi ; the maximum pressure
allowable during hydrostatic testing was 3,125 psi. The Technical Specification limit set by the NEC was 2,750 psig.
124
to try to inject [high pressure injection
water] to move the steam. (359)
It was only after the system reached high pres-
sure without collapsing the steam that some, but
not all, members of the emergency team con-
cluded there was superheated steam in the hotlegs.
Rogers stated,
. . . [We knew they had superheat] from
the time when they got up to normal
system pressure and had the resistance
bridge to the RTD's (resistance tempera-
ture detector) hooked up] . . . During that
period of time with high pressure we con-
cluded and . . . essentially we all agreed
we were at superheat. (360)
Miller also indicated repressurization led him
to recognize they had superheated steam in the
hotlegs :
. . . We were considering going higher
in pressure but by that time we had dis-
cussed steam conditions and going higher
wouldn't help us ... We were pumping . . .
as high a pressure as we had decided to
go. and the water level [was] not charg-
ing the system solid, and in fact we were
losing water to the reactor building
floor. 11171 In other words, very hot super-
heated conditions. (361)
Zewe also said he recognized superheat. Follow-
ing repressurization. he had consulted the steam
tables 11S in the control room from which the prop-
erties of steam in the hotlegs could be determined
and had concluded that superheated steam was
present. (362) However, he said that because
control room personnel realized there were steam
bubbles in the hotlegs. they were not sure that, the
hotleg temperature readings were correct. (363)
On the other hand. Logan and Ross said they
were not aware of the superheated conditions, Bug-
gating that Miller. Rogers and Zewe did not
share their conclusions with the whole team.
Logan said that such a condition was never
made evident to him. (364) Ross, who was directly
under Miller in operational control of the plant,
stated :
I don't think I ever personally put it
together and said. "Jesus, superheated
.-team." I knew we had a problem: we
couldn't fill the loops. I don't think I
ever inade it to. "Gee, we're super-
heated."
... I don't think I ever deduced anjthing
about superheated steam. (365)
This is a further indication that superheated
conditions were not discussed by the entire team.
Beyond that, there is no evidence that those who
recognized superheated steam in the system ever
discussed its origin, its consequences or its meaning
in terms of returning the plant to stable conditions.
Why Superheat Was Missed
The control room personnel gave a number of
reasons for their failure to analyze the implica-
tions of superheat. Zewe explained that he did not
know the true temperature of the core itself; he
was not aware of the earlier incore thermocouple
readings Porter had given Miller. (366) Further,
while both Zewe and Rogers said they realized
that the hotleg temperatures they were getting,
and which they considered reliable, were good in-
dicators of superheated steam in the hotlegs, their
statements indicate they also knew their hotleg
temperatures were not necessarily useful for un-
derstanding the properties of steam in the core.
(367)
Rogers explained :
Question : And you weren't relying on
the hotleg temperature as an indicator
of the reactor coolant system ?
ROGERS : You could not.
Question : Why couldn't you ?
ROGERS: Because there was no water
passing through the core getting to the
notice RTD [resistance temperature de-
tector]. (368)
In addition, unlike senior XRC officials later
in the afternoon, both Zewe and Rogers indicated
they undeistood that the core could be covered
even when temperature readings in the hotlegs
signified sujjerheated steam conditions in the
higher regions of the reactor.
Zewe was questioned on this point:
Question : But if you had the core cov-
ered would you have expected to have
seen that [superheated] condition in the
hotlegs?
ZEWE : . . . I think that it is conceivable
we could still have water in the core, but
still have steam voids in the hotleg be-
cause they were above the elevation of the
core, and we could have bubbles formed
high in the hotleg and have [water] in
the vessel ; yes. (369)
Rogers explained how he reached a similar
conclusion :
. . . Let me say something that helps
[explain] that. " In this plant, several
months previous to [the accident], dur-
"' Water was lost through the stuck-open PORV as the operators opened and closed the block valve to regulate
pressure.
'" See p. 107.
125
ing the [hot] functional testing program
when the core [was] not installed, a phe-
nomenon had occurred where we had
trapped a lot of hot water in the hotlegs,
and subsequently had the rest of the sys-
tem colder. And without the ability to run
the reactor coolant pumps, which we did
not have at that time, we could not get the
heat out of those hotlegs ; even with the
system filled with water, we could not
move any heat from that. It's in a natural
trapped condition. So this is not some-
thing that really startled me, that I had
hot conditions in the hotlegs and the rest
of the system lower temperatures. That
was not something new to us ... It was
accepted as a condition because of the
layout of the plant ... It has happened at
other B&W plants, so it was not a brand
new problem. 119 (370)
Neither explanation, however, addresses why
they did not discuss past uncovering as the source
of the superheated conditions.
Unlike Zewe, Miller did know of some of the
incore thermocouple readings, but had discounted
them as unreliable. 120 He said he was relying on
hotleg temperatures, which he believed were the
hottest temperatures in the system :
. . . The resistance temperature detector I
have spoken of was reading around 720
degrees. The cold temperature was read-
ing I think, less than 200. The steam [gen-
erator] downcomer temperature was
around 500 or so at various times; the
RTD [resistance temperature detector]
in the hotleg being the hottest ... At that
time I would probably assume the core
was somewhere below . . . 700 without
knowledge of specifically why. (371)
At the same time, Rogers had incorrectly inter-
preted the pressurizer temperature as the best in-
dicator of core temperature. He believed it was a
better indicator of core temperatures than the
hotlegs :
... I accepted that the temperature in the
pressurizer was a result of the water com-
ing in through the core and going out
through the surge line to the pressurizer.
I accepted [it] myself, and a lot of other
people were accepting the same
thing 121 (372)
None of the other control room personnel inter-
viewed by the Special Investigation staff men-
tioned this position. Miller, for example, implied
that pressurizer instrumentation as a whole could
not be relied on :
. . . We really didn't trust the pressurizer
instrumentation because we knew the con-
dition of the loop being [having] steam
bubbles. (374)
Rogers also told Special Investigation staff that
he was unaware of any steam condition in the core
itself, either during repressurization or at any
other time in the day. Nor, Rogers said, was it
likely he or others would have deduced that :
. . . No, I would not say that I knew that
we had [steam and water] conditions in
the core. We were totally convinced we
had steam in the loops, and I believe with
what we knew we were putting in there,
we would not have assumed that we had
steam in the core at that point in time,
and probably wouldn't through that day
with the information we had anyway.
(375)
Both Miller and Rogers gave as a reason for
their not recognizing core uncovering as the source
of superheated steam their preoccupation with
identifying a strategy for returning the plant to
stability :
MILLER: The only action I considered
was to maintain core coverage. I don't
think I went back and asked myself how
we had gotten to this point . . . (376)
* * *
Question: What led to superheat?
ROGERS : ... in the control room at that
point in time and probably throughout
the rest of the day we didn't have the
luxury nor the need to go back and look
at. how we got to where we were. (377)
This failure by the utility to establish trends or
to look back at causes of conditions was a serious
analytical weakness that would be repeated again
during the day. It was an obstacle to an effective
response by outside agencies, in particular the
NRC.
Differing Recollections
Based on the collective recollections of Kunder,
Logan and Ross, concern about a steam bubble
119 This design problem was recognized by GPU and B&W during hot functional testing of the plant, but apparently
was not communicated to operators prior to the accident. See "Prior to the Accident," p. 63.
120 See pp. 112-113.
"' In fact, pressurizer temperature was not a good indicator. The evidence suggests that pressurizer temperatures
during the accident were hundreds of degrees below actual core temperatures. The relief valves on the pressurizer, as
well as the loop seal and the isolation of the temperature measuring devices from the pressurizer vessel, would tend to
prevent these devices from registering superheat. Other evidence suggests that Rogers was led to focus on pressurizer
temperature as a result of a recommendation from a Babcock & Wilcox task force in T.ynchburg, Virginia. (373)
126
on top of the core and uncertainty over whether
the core was uncovered was one of the motivating
factors in the next step they would take depres-
Miri/.ation. Logan said. "The determination was
made that we . . . had a bubble in the top of the
IJo: recalled that this determina-
tion was made during the planning for depres-
surization. (379) Yet Miller, who recalled a con-
cern that flow from HPI was bypassing the core,
- I did not recall any discussion of steam in
the. core :
Question : Were you confident there
wasn't steam in the core '.
MII.I.ER: I don't believe we were con-
fident of that, but I can't honestly re-
member discussing that . . . We talked
about heat removability, I think, mow
than steam bubbles in the core. . . . (381)
In hindsight, there was an unwillingness among
mine control room personnel to believe the worst,
and inadequate communication among those
present.
Two weeks after the accident Miller stated that
during that first day he "did not admit the reality
of failed fuel." (382) Flint has said repeatedly
that he told Rogers that morning that the core
probably had been uncovered. Rogers, however.
>aid he did not recall hearing that significant
information.
Similarly, when Flint expressed his doubts
about repressurization. given his belief that the
.-team in the hotlegs was superheated, he said
his advice was not heeded. (383) He was proven
correct, but that did not lead to a general aware-
of that condition. Xuclear engineers in the
plant also had concluded there had been voiding
in the core, but members of the emergency com-
mand team have not indicated they were aware
of that information. Similarly, there is no evi-
dence showing that the opinions of the instrument
technicians, who took the incore thermocouple
readings and from them deduced core uncovering,
ever reached the team. Thus, as the alxrve dis-
cu>sion illustrate:-, there were both physical indi-
cations of the condition of the core and several
cleai- statements about those indications, yet many
members of the utility's emergency command
team apparently remained unaware of them.
It is also important to recall that some two hours
prior to depressurization, Edson Case of the XRC
had surmised on his own that part of the core had
been uncovered and had expressed this opinion to
Commissioner Aheame. (384) Yet neither the
Commissioners nor the XRC emergency response
staff pursued this concern directly with the utility
or discussed the need for protective action with
the State. 122
FAILURES IN COMMUNICATIONS
At 9 :'2fi a.m.. during the time that the utility was
repressurizing. Kunder spoke on the telephone
with Donald Haverkamp at Region I. 123 Haver-
kamp asked Kunder to "go through the scenario'"
of what had happened earlier in the morning.
O.'i) During the conversation, Region I and.
through it. IRACT in Washington received some
important information.
Kunder told the regional office that the reactor
coolant pumps were off and that there was no flow
through the primary system. Part of Kunder's ex-
planation went as follows:
. . . The pressure came ... all the way
down to about 1.000 pounds and that was
roughly over a 15 minute span. I think it
was during that condition that we . . . got
a bubble [or] some such through appar-
ently the heating in the core up in the
loops and ... it apparently had an effect
of vapor locking ... It looks to me
[like] we had that vapor locking effect
being fed by the heat in the core . . . The
problem [then was] trying to get the pres-
sure down low enough so we are sure that
the flow is going down into the reactor
vessel annulus n - J ' and up into the core.
Yai>or lock is apparently preventing that
from occurring. (386)
Region I did not report to IRACT that there
were steam conditions in the core at that time. This
was a serious breakdown in communications that
adversely affected the XRC's potential ability to
understand the accident. However. Kunder did not
specifically tell Haverkamp there was steam in the
core: he mentioned only that vapor locking was
causing the problems with flow.
When Special Investigation staff read him parts
of his conversation with Haverkamp. Kunder re-
sponded :
. . . That's interesting. That's more accu-
rate than I have been recalling. My per-
ception at that time. I think, was. I guess.
pretty accurate in the sense that I was
aware we had steam in the core and in the
hotlegs. los?)
15 minutes after IRACT learned that the
reactor coolant pumps were off and there was no
flow. Region I Director (trier called John Davis.
Davis asked whether, given this condition, Grier
'" See the discussion of evacuation, pp. 132-135.
1:4 Fifteen minutes earlier Region I had begun tai>e recording communications between the site and its office. Thus.
there was a record of the conversation.
;i The space between the reactor vessel wall and the core.
127
had any concerns about adequate cooling of the
core. He replied, "not as long as pressure and tem-
perature continue to come down." (388)
At that very moment the utility was trying to
repressurize to increase pressure in the primary
system, not bring it down. (389) Grier's statement,
viewed in the context of what was actually hap-
pening at the plant, typified the communications
problems that hampered the NRC's response
throughout the morning and early afternoon.
WHAT THE NRC HAD LEARNED
By 9 :30 a.m., IRACT had received the following
information from the region :
There was a failure of nuclear fuel. (390)
There was high radiation in the contain-
ment. (391)
There was increased containment pressure.
(392)
The reactor coolant pumps had been shut
off. (393)
There was boiling water in the reactor cool-
ant system. (394)
These facts indicated that the plant had had sig-
nificant problems, was still in an abnormal condi-
tion and had, earlier in the morning, experienced
some problem with cooling the core. This was, in
fact, the essence of Edson Case's interpretation of
the situation just before 9 a.m. (395) However,
NEC headquarters lacked the data the utility and
its own Region I personnel had that provided a
clearer indication of core uncovering. (396)
THE NEXT STEP: DEPRESSURIZATION
Between 9 :45 and 11 :30 a.m., the emergency
command team decided to depressurize the reactor.
(397)
A number of concerns prompted this strategy.
Because attempts to achieve natural circulation
had failed, control room personnel had been using
the make-up pumps to provide water to cool the,
core. Several said they were becoming worried the
supply of cooling water would become depleted.
According to Ross,
. . . We were getting concerned that we
were going to run out of water soon . . .
[Depressurization] was kind of a rash
move, we felt at the time. But we felt it
was necessary. . . . (398)
Miller concurred,
... I was concerned with the amount of
water. We had hours of water left, but
were talking about bringing in the ulti-
mate water sources at the time. (399)
Another reason control room personnel gave for
depressurization was that repeated operation of
the block valve, used to regulate pressure during
repressurization, might cause it to fail. (400)
The most significant reason, however, was uncer-
tainty whether the core was uncovered. The dis-
cussion regarding depressurization did not focus
on superheat, but rather on the possibility that a
saturated steam bubble atop the core might be in-
hibiting flow through the core. (401)
As Miller recalled,
. . . We knew there were steam bubbles
within some of the pipes. We looked at
elevation diagrams, and I remember some
of that kind of analysis. There were peo-
ple in the group throughout the morning
who postulated that the high pressure in-
jection [of coolant water] possibly could
be by-passing some of the core. (402)
Rogers had the same recollection.
... At some point in time in the meeting
which preceded our reducing the pressure
again, the question was brought up, "Are
we absolutely sure that the high pressure
injection water is getting to the core?"
"Are we absolutely sure that the core is in-
deed being covered?" (403)
Following the accident, Kunder recalled a con-
cern of his that chemicals in the coolant water
might be concentrating in the core and blocking
the passage of water:
... I know another thing that I was wor-
ried about, and I think we all shared the
same concern : were we
boric acid in there ? (404)
Not all the control room personnel shared the
various concerns. Miller, for example, responded
to Kunder's statement about the concentration of
boron by saying "that never bothered me at that
time." (405)
The weight of the evidence suggests that the.
primary concern was the possibility that the
core was uncovered. 125
concentrating
Ihe XRC s Special Inquiry Group stated : "At 11 :30 a.m., because no one can think of anything else that has not
been tried, the decision is made to depressurize the system. Later, it will appear that there are several different
perceptions of the precise reasons for depressurizing." (406)
Although parts of this conclusion are correct, on the whole it is misleading. In contrast with this Investigation,
the Special Inquiry Group did not point out that there were two separate attempts to depressurize, each with separate
128
INTERPRETING THE CORE FLOOD TANKS
Some control room personnel said they were
looking for a way to assure themselves the core
wa> covered. They had no means of measuring
water level directly, and they had long since
discounted pressurizer level as a reliable in-
direct indicator. They reasoned that by depres-
surizing they could force the core flood tanks
to come on and inject water onto the core.
Several members of the management team, in
interviews with Special Investigation staff, de-
scribed how they expected to be able to in-
terpret the behavior of the core flood tanks.
MILLER: ... if the core was signif-
icantly dry and pressure differential ex-
isted, we felt we would push a lot of
water into the core. (408)
Ross: We assumed that if the core
was. in fact. very, very empty, and we
lowered pressure, that the core flood
tanks would inject, and cover the core
again. (409)
If. on the other hand, the water level was al-
ready high, they believed the tanks would inject
only a small amount of water, meaning that the
core was covered.
Miller, though he said he did not agree with
all the concerns, decided to depressurize :
. . . During that morning the group
that I assembled were discussing the fact
that we wanted . . . double assurance
that the water we were pumping in was
covering the core ... I didn't believe
the core was uncovered but I listened to
people in my group [who were] looking
for double assurance. (410)
DEPRESSURIZATION INITIATED
At about 11 :30 a.m.. the operators began to de-
pressurize by opening the block valve and de-
creasing HPL One hour later, at about 12 :41 p.m.,
the pressure at which the core flood tanks are ac-
tivated 600 psi was reached, and they injected
water into the core. (411) This continued for 30
minutes. Pressure continued to fall to about 435
psi, just above the 430 psi level that the NRC had
been told was the point at which the decay heat
removal system would come on. 126 (413) Only a
minimal amount of water was dumped onto the
core before the tanks shut off approximately 750
gallons equivalent in volume to only a minute
and a half of flow from one HPI pump. 127 (414)
DEPRESSURIZATION TERMINATED
At 1 :10 p.m., the operators closed the block
valve, thereby terminating depressurization. The
control room personnel said they were convinced
that the minimal amount of water injected into
the core meant it was covered. Miller, for one.
commented :
. . . We had a foot, foot and a half, of
water decrease [in the core flood tanks]
and that convinced most people in the
group that we didn't have major uncov-
erage at that point. (415)
Rogers noted :
. . . That [core flood tank injection]
occurred in a way that everyone agreed
was an indication of water in the core
. . . that was conclusive evidence that
there was a water phase in the core know-
ing full well they had a steam-noncon-
densible phase in the hotlegs. (416)
According to Rogers, he and some other control
room personnel went a step further and concluded
that, since the tanks did not dump all their water
at once (an event that clearly would have indi-
cated that the core was uncovered at the time), it
had never been uncovered at all. As Rogers told
the Special Investigation staff:
... I will readily admit that that
statement [that the core had never been
uncovered] was made quite a few times
in the control room at that point in time
when we deliberately went down to the
core flood tank float condition, the ob-
jective being assuring the core is covered.
When that occurred more than one per-
son in the control room said. "Hey, the
objectives. The purpose of the first depressurization. when it was planned, was to ensure the core was covered, not to
bring into operation the low pressure decay heat removal system. As Rogers told Special Investigation staff:
. There was no deliberate action to go to low pressure injection conditions at that point. It was agreed that
at some i*Ant in time we will want to get there . . . [But] the concern over whether the core was covered is
why that action was taken . . . We would want to get the water phase (in the hotlegs) before we [went] to
decay heat. (407)
The objective of the second depressurization to use the low pressure decay heat removal system was stated in
a conversation between Miller and Ross : the content of their discussion was not generally known, even to control room
Iiersonnel. until two weeks after the accident. This would explain the differing recollections of control room personnel
about the purpose of depressurization.
"* In fact, the setpoint was 320 psi. (412)
^ It is unclear why so little water was released from the core flood tanks. It could be that the core was covered at
that point, or it could have been the result of the effect of steam and gas in the reactor coolant system. See pp. 142-143.
129
core's covered and it probably has never
been uncovered." (417)
It is clear, in hindsight, that they were wrong.
Furthermore, there is no evidence that the com-
mand team took into account how the core flood
tanks would behave in a situation where the core
was only partially covered or being cooled by
steam. In both cases, depressurization would tend
to increase boiling, since water boils sooner at
lower pressures. The boiling would in turn in-
crease the volume of steam in the system, which
would in turn raise pressure in the system. This
would inhibit the flow of water from the core flood
tanks, as they empty only if pressure in the reac-
tor vessel is lower than in the tanks. Thus, with
a core that was partially uncovered, the core flood
tanks would only be able to discharge a portion of
their contents, since the injected water would
quickly flash to steam, raising pressure and there-
by reducing the pressure differential between the
core flood tanks and the reactor vessel.
Further, the NRC has pointed out in an early
investigation of the accident, that because of the
layout of the core flood tank piping, minimal in-
jection by the tanks cannot be interpreted to mean
the core is covered. (418) When there are satu-
rated or superheated steam conditions in the sys-
tem, the core flood tanks can become cut off from
the core. The steam can form loop seals in the core
flood tank piping which can prevent the tanks
from injecting water onto the core even if the core
is totallv devoid of water. (419)
HPI THROTTLED AGAIN
Shortly after 1 :10 p.m., a control room opera-
tor's log showed the entry : "Stopped HPI." (420)
While the meaning of this entry has never been
explained, the evidence shows that over the next
several hours the amount of water being added to
the system was again throttled to low rates of
flow. (421)
THE NRC ONSITE
A five-member NRC inspection team arrived
onsite at 10 :10 a.m.. when radiation in the Unit 2
control room had reached levels that required
evacuation of non-essential personnel. Personnel
were moved either to the Unit 1 control room,
where the Emergency Control Station was being
transferred, or to the Three Mile Island Observa-
tion Center, a facility for visitors adjacent to the
plant on the east bank of the Susquehanna. Those
who remained in the Unit 2 control room had to
use respirators periodically for the rest of the day.
(422)
Miller directed the NRC inspectors to the con-
trol room at Unit 1. (423) The NRC team had
been sent by emergency vehicle at 8 :45 a.m. from
the Region I office near Philadelphia, about 85
miles away. 128 It was comprised mainly of health
physicists; there was only one operations in-
spector. (425)
According to the Region I plan, the Project
Inspector for the facility "normally" is to be a
member of the inspection team and to serve as
its leader. (426) However, according to Brunner.
because of the erroneous information about radi-
ation releases that morning, the Region thought
the problem was largely radiological. (427) The
TMI Project Inspector therefore was not in-
cluded on the team. (428) Instead, the Region
decided to monitor the operational problems by
telephone from the Regional Incident Response
Center. (429)
Once again, misinformation had been trans-
mitted and resulted in an inappropriate response.
Further, as will be seen, the presence of NRC
representatives onsite did little to improve the
flow of communications or the accuracy of the
information that was transmitted.
Shortly after the inspection team was dis-
patched, a two-man back-up team with a second
operations inspector left for the site. Later in the
afternoon the Region I Project Section Chief for
TMI was sent. (430)
INADEQUATE REGIONAL PROCEDURES
The NRC's overall goal is to protect the health
and safety of the public. To meet that goal, it is
essential that the NRC have the ability to respond
quickly in the event of an accident. This basic
premise was a lesson the NRC had learned from
its analysis of its inadequate response to the
Browns Ferry fire of 1975. (431)
Despite that overall goal, the procedures in the
Region I plan allowed up to 6 hours before an
NRC team had to be onsite, even in the case of
the most serious category of nuclear accident.
(432) There was, however, another procedure
Region I could have followed, given the condi-
tions of the TMI accident. This procedure pro-
vided for the dispatch of the inspection team by
two predesignated modes of transportation both
by emergency vehicle and by either helicopter or
chartered aircraft. (433) Region I decided to use
an emergency vehicle only, (434) and although
the reactor is located only about 85 miles from
28 The Plan specifies that emergency vehicle and "rotary or fixed wing aircraft" transportation are appropriate in
the case of a Level I incident (defined as "an event which has an actual or imminent serious threat to public health and
safety, the environment, property, or security and safeguards of licensed facilities and materials) under certain condi-
tions." Those conditions relate to weather, time of day and distance from the regional office. (424)
130
the regional office, the inspection team did not
arrive until nearly two and a half hours after
notification. (435)
In correspondence with Special Investigation
staff. Boyce Grier. Director of Region I, noted
that the second procedure relating to transporta-
tion of the inspection team was not a requirement
but merely "guidance on selection of the trans-
portation "mode." (436) He also wrote:
. . . Since the team departed Region I
at 8 :45 a.m. [one hour after notification]
and arrived at the TMI-2 control room
at 10:05 a.m. [2:20 after notification]
the intent of the Plan was met, (437)
In fact, the onsite inspection team did not set up
operations in the Unit -2 control room until about
11:30 a.m.. nearly an hour-an-a-half after their
arrival at the plant during the evacuation of the
Unit 2 control room.
Had Region I used air transportation to the
site, along with the simultaneous dispatch of an
emergency vehicle, the XRC could have mitigated
the communications difficulties that were a direct
result of the team's arrival during evacuation of
the Unit 2 control room.
MORE COMMUNICATIONS PROBLEMS
When non-essential personnel were evacuated
to the Unit 1 control room, the original phone con-
nection between the Unit 2 control room and the
Region was broken since Warren, the individual
manning the phone there, was among those
evacuated.
Although a line was established between Region
1 and the Unit 1 control room shortly after the
arrival of the XRC on?ite inspection team, for a
period of about an hour-and-a-half there was no
direct communication between the Unit 2 control
room and either the Region or XRC headquart-
' This was a critical period, a time when
there was uncertainty about whether the core was
covered, when a new strategy was being planned,
and when the dimensions of the accident and what
the utility intended to do or was doing to achieve
stability should have been communicated directly
to offsite agencies.
It was during this period of no direct contact
with Unit 2 that the Commissioners asked for
hourly briefings from the EMT. unless a significant
development necessitated a more immediate call.
(439)
Prognosis at NRC Headquarters
At about 10:16 a.m.. the Commissioners at head-
quarters Gilinsky. Kennedy and Bradford re-
ceived the first collective briefing from EMT mem-
bers Davis. Case and Lee Gossick. With respect to
the status of the reactor, Case told the
Commissioners :
I think right now we have the situation
under control and we'll have to keep get-
ting information to make sure that con-
tinues. (440)
After this briefing. Commissioner Bradford
decided to join Commissioner Ahearne at the Re-
sponse Center in Bethesda. This meant that there
were two Commissioners at the Response Center,
one at a local hospital and two at XRC headquar-
ters in Washington. D.C.
Others in IRACT and the EMT were conveying
the same favorable information about the status
of the reactor. At 10:32 a.m.. IRACT's Brian
Grimes briefed the Director of the Office of Xuclear
Reactor Regulation. Harold Denton. Grimes told
him that as far as he could tell, the reactor's core
was in "a fairly normal status." although it was
not clear to IRACT that the operators had reestab-
lished the proper water level in the pressurizer.
(441)
During the next briefing of the Commis-
sioners, around 11 a.m., Gilinsky asked whether the
reactor was "under control." (442) Case told him :
The signs are encouraging, Vic. Pres-
surizer level is up ... we have coldleg
temperature measurements of 220. which
is good. We have indications that one
steam generator is being used for nat-
ural circulation and transferring heat
outside the primary system. So signs
. . . continue to be good. 129 (443)
IRACT member Harold Thornburg made
similar statements when briefing the Regional
Directors of the Office of Inspection and En-
forcement at about the same time : "They're
getting the situation under control/' (444) "it
looks like the damn thing seems cool now."
(445) "it does look like thev are cooling the
thing." (446)
The Flow of Misinformation
Part of the reason for the transmittal of in-
accurate information was that IRACT was get-
ting the wrong information about a kev con-
dition natural circulation. This was contributing
to their incorrect analysis of the accident. In addi-
tion, they were not getting the proper data on hot-
leg temperatures. Following repressurization at
9 :30 a.m., hotleg temperatures in the "A* 7 loop had
increased from about 680 F to 730-740 F. where
they remained until about 11 :30. by which time the
utility had reversed its strategy and was attempt-
ing to reach stability through depressurization.
( 447) There is no evidence that XRC headquarters
was made aware of this upward trend in tempera-
' On the accuracy of this account, see p. 132.
131
tures. In fact, for two-and-a-half hours TRACT
continued to receive only the hotleg-coldleg aver-
age temperatures, mistakenly reported either as
hotleg or as primary system temperatures. (448)
Misleading conversations between Met Ed em-
ployees and NRC officials in Region I con-
tributed to the confusion between average tem-
peratures and hotleg temperatures. Typical was
an earlier conversation at about 10 :15 a.m. be-
tween Kunder in the Unit 2 control room and
Region I's Haverkamp.
Kunder told Haverkamp that the unit super-
intendent in charge of operations, Mike Ross, was
certain the core was covered. (449) It is now evi-
dent that some control room personnel were, at
that time, concerned whether the core really was
covered; it was this concern that had led to the
decision to depressurize. 13 Kunder did not men-
tion any of these doubts concerning core coverage.
Although he did convey the control room person-
nel's doubts about the hotleg temperatures, in
doing so, he mistakenly characterized the average
temperature reading of 571F as the hotleg tem-
perature :
KUNDER (to Mike Ross in Unit 2 con-
trol room) : Mike, how does the core look ?
KUXDER (to Haverkamp) : [I'm] talk-
ing to Mike Ross he's looking at the in-
dications; his assessment is that he's
surely . . . got the core covered and we are
getting water . . . into the core. The only
thing though is that the T h [hotleg tem-
peratures] are still high and that's what
bothers us ; the pressure, and getting con-
trol of it, and . . .
HAVERKAMP : What is your pressure and
temperature now ?
KUNDER: The pressure is still up
around what I told you, it's holding there,
okay : We got a bubble in the pressur-
izer . . . But he is still baffled by the T hot
[hotleg temperatures] ; we are really try-
ing to access that. T hot right now is
reading 571 degrees F but, again, I am
not sure how real a number that is. (450)
At about 11 :40 a.m., IRACT received both hot
and coldleg temperature readings. The Region re-
ported the hotleg temperature to be 620F. (451)
It did not tell IRACT that 620F was not a real
measure of temperature in the primary system, but
merely the upper limit on the scale on the control
room console and that the instrument was reading
offscale high.
The actual temperature at that point was about
730F. While the control room personnel knew the
reading of 620F from the strip chart on the con-
sole in the control room was inaccurate, there is
no evidence that the NRC inspectors who were in
the control room or that IRACT in Bethesda knew
this until 4 p.m.
The Meaning of Coldleg Temperatures?
IRACT was, at this time, getting accurate cold-
leg temperatures in the neighborhood of 220 F.
However, IRACT and EMT members interpreted
the accuracy of these readings differently. Dur-
ing the 11 a.m. briefing of the Commissioners,
Edson Case of the EMT reported the coldleg tem-
perature without mentioning hotleg temperatures
or the difficulty in acquiring them. He also told
the Commissioners he believed the coldleg read-
ing was "good." (452) Twenty minutes later, in
a briefing of NRC Region IV, IRACT's Harold
Thornburg stated that the coldleg temperature did
not "look right." (453)
EVACUATION
RADIOLOGICAL DATA
Transcripts of telephone conversations between
Region I and IRACT show that at 9 :08 a,m. the
Region informed IRACT that a Met Ed radiation
survey team had reported measurable levels of
radioactive iodine at the plant's perimeter. (454)
The survey team was said to have detected the
iodine while measuring a radioactive plume re-
leased from the plant. James Sniezek, who was at
IRACT that morning assembling and analyzing
incoming radiation measurements, concurred with
Special Investigation staff that a reading of ra-
dioactive iodine from a plume at the site's perim-
eter would indicate a subsequent offsite release.
(455) Indeed, a 9:22 a.m. air sample taken in
Goldsboro on the west bank of the Susquehanna
River opposite the site was reported to contain
measurable levels of radioactive iodine. (456)
Although IRACT learned about the radioactive
iodine in the plume over an hour before the NRC
issued its first press release at 10 :30 a.m., the NRC
informed the public in its press release that :
Measurements are still being made to de-
termine if there has been any radioactiv-
ity detected off the site. There is no indi-
cation of a release off the site. (457)
Joseph Fouchard, Director of NRC's Office of
Public Affairs, was with the EMT throughout
March 28. Fouchard said that the information in
the NRC's press releases was based on conversa-
tions he had with members of the EMT. (458) As
noted, the EMT was receiving reports on site
1 See pp. 128-129.
132
Personnel plot wind directions to assist in radiation monitoring
status and radiation from IRACT. Fouchard did
not recall clearing the press releases with Sniezek
at IRACT. but Sniezek indicated IRACT person-
nel saw extracts of them. (459)
Inaccurate radiological information was trans-
mitted at other times on Wednesday and dissemi-
nated to the public, reflecting the inadequacy of
both the radiation monitoring around the site and
the communications system set up by the XRC to
receive data. But, as is discussed below, these prob-
lems only partially explain why so little attention
was paid to the possible need for protective action
in the earliest hours of the accident.
RESPONSIBILITY FOR EVACUATION?
At 11 :09 a.m., the EMT reported to the Com-
missioners by telephone that radioactive iodine
had been detected offsite and that the sample was
going to be analyzed. 131 Gilinsky. as he would dur-
ing subsequent briefings, questioned who was re-
sponsible for evacuation of the public. (460) This
important issue had not been addressed seriously
by any group up to this point, despite the recog-
nized uncertainty over core covering.
Davis explained :
... of course, the licensee will analyze
them. We will see what the levels are.
They'll flow in here, and at some point,
as they begin to increase, they would
move into emergency measures such as
evacuation. But we are a long way from
that from what we've got now . . . And
that would be through the state.
GILIXSKY: But I want to understand
who has responsibility here ?
DAVIS : Okay. That's through the state.
GILIXSKY : So it is the licensee dealing
through the state at this point. (461)
It is clear from this conversation that on the
first day of the accident, the EMT did not believe
m This was the same iodine release Domsife reported. See p. 135.
133
the NRC had any role to play in determining the
need for evacuation. 132
During a later briefing, at 1 :45 p.m., Gilinsky
would learn for the first time of onsite readings
of radioactive iodine above minimum detectable
levels. He again asked what levels of radioactivity
would trigger mandatory protective action initia-
tives :
GILINSKY: ... At what sort of levels
do you begin to start talking about mov-
ing anybody from where they are.
DAVIS (to Brian Grimes in back-
ground) : At what sort of levels do you
begin to evacuate ?
GRIMES : Evacuate ? Thousands of times
higher.
GOSSICK: Thousands of times higher
than what we're getting now.
GILINSKY : I see. Okay.
DAVIS : 5 Eem whole body, 25 Rem thy-
roid type numbers. 133 (464)
PROCEDURES ARE NOT UNDERSTOOD
The above and subsequent conversations reveal
a serious deficiency in the NRC's emergency re-
sponse : its limited knowledge concerning respon-
sibility for protective action and the correct pro-
cedures to be followed in determining its need.
Two points become clear from the questions and
answers. First, the Commissioners had little under-
standing of who was responsible for recommend-
ing or ordering evacuation. Neither they nor EMT
saw the Commission as having a role. The Com-
missioners expressed no awareness of the new
guidelines EPA was in the process of issuing with
respect to procedures for estimating projected
doses and evaluating the need for evacuation. 134
Rather, both they and the EMT were considering
the need for protective action only in terms of the
relation between the Protective Action Guide
(PAG) dose the "projected dose to individuals in
the population which warrants taking protective
action" and an actual offsite radiation reading.
This meant, in effect, that they would only have
considered evacuation in the event of an actual re-
lease of radiation at or above a certain level speci-
fied by the PAG. Neither at this time nor at any
subsequent time did the Commissioners or the
EMT use as criteria for evacuation reactor system
status, present or future.
The EPA made clear that PAGs were to be used
"only in an ex post facto effort" 1 * to minimize the
risk from an event which is occurring or has al-
ready occurred and that under no circumstances
were they to be considered as "acceptable doses."
(466) In fact, according to the EPA, 134 "there is no
direct relationship between acceptable levels of
societal risk and Protective Action Guides." (4f>7)
The EPA specifically recommended that PAGs
be applied, and decisions of protective action
reached, in conjunction with ongoing estimations
of projected doses. (468) The latest EPA guid-
ance stipulated that these projected doses were to
be determined on the basis of specific information,
including "reactor system status" or "plant con-
ditions." (469) Implicit in the use of this criterion,
and explicitly stated elsewhere in the Manual, is
that protective actions such as evacuation can and
obviously should be taken before the hazard is
already present. (470)
Gilinsky has since acknowledged that the NRC
did not clearly understand the EPA guidelines on
protective action :
Question : Was the executive manage-
ment team, as far as you could determine
based on your conversations with them on
March 28, in a mode in which the question
of evacuation was, one, [solely] a matter
of state responsibility; and, two, to be
determined on the basis of the EPA
[Protective Action] Guides?
GILINSKY: I think the answer to the
first part is yes. I don't know that every-
one was clear on the EPA guidelines. In
fact, I think the answer is they were not.
Some people certainly were because there
had been a joint task force with EPA in
which they participated. (471)
A VACUUM IN RESPONSIBILITY
That there was a vacuum in responsibility at
this time concerning determination of the need for
13S On Friday, the NRC recommended evacuation. Significantly, when that recommendation was made, it stemmed
from a single measure of radiation released from the plant, a release (1,200 mR/hr) that was actually smaller than one
on Thursday afternoon (3,000 mR/hr). IRACT personnel were aware of Thursday's reading, but apparently the EMT
was not because it made no recommendation for evacuation at that time. On Friday, EMT officials recommended evacua-
tion without consulting their support staff at IRACT. Friday's recommendation had no relation to EPA guidelines on
projected doses ; it was based on a one-time-only reading from a puff release monitored by a health physicist in a helicopter
hovering directly above the plant's vent stack. (462)
33 These PAG dose rate levels are maximum levels at which protective action is mandatory. However, the PAGs also
specify that lower range levels (1 rem whole body and 5 rem thyroid) are applicable if there are no local constraints on
protective action. Moreover, even lower levels pertain for "sensitive populations" such as women of childbearing age and
children. (463)
M The EPA had held numerous meetings and discussions on the proposed revised guidelines. The NRC participated
on at least one joint task force relating to them. (465) Therefore, the NRC should have had some familiarity with the
new guidelines.
134
protective action is apparent. The XRC saw itself
as having no role: it assumed that that function
was the State'?. Those people within the State re-
sponsible for protective action said they saw no
reason seriously to consider the matter throughout
the day. based on their comparison of the EPA
Protective Action Guides and the data on offsite
releases they had available. (472) The State was
not aware of the new EPA guidelines, nor was it
ing plant status on its own in order to formu-
late protective action. 1 ' 3 Moreover, throughout the
clay, it received assurances from the site that the
plant was stable. Dornsife never questioned
whether the reactor core was covered. Based upon
this belief and the assurances from the utility, the
BRP told the Governor that no protective action
was necer-sary. (473)
Had Dornsife known of the uncertainty of util-
ity personnel and the XRC officials about the core
being uncovered, he said he would have asked more
operational questions, since he believed that ex-
tended uncovering of the core was a reason to con-
sider evacuation. (474)
For it? part, the utility was monitoring onsite
and offsite levels of radioactivity and focused too
heavily on them as its criteria for evacuation,
rather than on plant conditions, (475) It should be
noted, however, that the utility was confused as to
actual plant conditions. It is unclear whether the
utility would have considered its uncertainty a
plant condition to be used as a criterion in consid-
ering the need for protective action.
The lack of understanding on the utility ? s part
mplified by Dubiel's misreading of a signifi-
cant indicator of plant conditions the wide dif-
ferential in temperatures between the hot and cold-
1 Dubiel was in charge of supervising the
utility's implementation of the TMI Emergency
Plan with respect to radiation protection. Yet he
told Special Investigation staff he was not "overly
concerned or worried ibout that condition." (476)
Similarly, he believed (hat because the core flood
tanks, when activated that afternoon, had injected
only a limited amount of water, the core was
covered. 1 - 7 (477)
COMMUNICATIONS WITH THE STATE
The difficulties in communications between the
State and the utility were the result of an inade-
quate number of technically qualified State per-
sonnel and the utility's deficiencies in transmitting
information, as discussed below.
DISJOINTED FLOW OF INFORMATION
A typical example of the disjointed flow of in-
formation was Lt. Governor Scranton's 10 :55 a.m.
press conference. The State was receiving the
same kind of misleading information as the XRC.
and Scranton expressed optimism about plant
conditions, just as the XRC had been. He read a
press release that Paul Critchlow. the Governor's
Press Secretary, and David Milne. Department of
Environmental Resources' Press Secretary, had
helped prepare earlier that morning. It stated that
"no increase in normal radiation levels have been
detected." (478) He had been told that at an ear-
lier briefing; (479) the information was based on
what Miller had told Dornsife earlier in the morn-
ing. 118 (480)
After the statement was read, Dornsife was
called upon to answer technical questions from the
press. (481) Just, before the press conference, at
about the same time the Commissioners learned of
the release, Dornsife had heard from Thomas
Gerusky. the Director of BRP. that detectable
levels of radioactive iodine had been found in
Goldsboro. Dornsife had not had an opportunity
to tell the Lt. Governor and other State officials
before the press conference, but he announced the
release at this time. Dornsife said the Lt. Gover-
nor and his staff were both surprised and discon-
certed. (482)
PERSONNEL AND COMMUNICATIONS
Several things contributed to the communica-
tions difficulties between the utility and the State.
For one, the State did not have enough technical
people who were capable of collecting, coordinat-
ing and disseminating radiation information on a
24-hour basis. (483) When TMI wanted to relay
field survey data to BRP. often the only person
available to take the information was the BRP
secretary. (484) The utility personnel who had
been evacuated from Unit 2 to Unit 1 said that
having to give technical data to a lay person im-
peded the efficient flow of radiation information to
BRP. According to Benson,
Question : When you were talking with
the State, who were you speaking with *.
BEXSOX : I don't recall. Sometimes I
felt the person at the State was not too
educated in engineering fields. I felt it
was a secretary, because she would ask
me. was that "gamma." It was really just
"* See p. 134 for the implications of this condition.
m See p. 106.
117 See p. 130.
"* Before going to the Capitol Building to brief Scranton, Dornsife had called Miller for more detailed information.
Miller explained to Dornsife that the high radiation readings reflected "gap activity" caused by a low-level pressure
transient. ( A gap exists between the fuel jiellets and the Zircaloy cladding surrounding them. Gaseous fission products
tend to collect in this space. Once the cladding has been breached 'failed fuel" these gases are released into the
coolant.)
135
some of their reactions ... I don't feel
the person was up to par on health physics
and that sort of thing, more like a secre-
tary. (485)
As pointed out earlier, Dornsife was the sole nu-
clear engineer in the State's emergency manage-
ment structure. 139 He had to handle several re-
sponsibilities. Among them, he was supposed to
assist Margaret Reilly, in charge of environmental
matters at BRP, with the radiation monitoring
program. BRP's limited resources were being
overwhelmed by the sudden influx of radiation
data. (487) However, Dornsife also was called on
to assist at briefings and press conferences. Ac-
cordingly, he was frequently called away from his
office and had trouble keeping current. As he
noted,
Sometimes Tom [Gerusky] and I were
off at meetings, and it was difficult to
after being away from the office to come
back and get current with what was going
on which led to the problem of keeping
current with what the plant status was.
(488)
Reilly said she found the comings and goings
troublesome, especially because she needed their
assistance in handling the radiation tasks for
which she was responsible :
. . . Dornsife and Gerusky spent a lot of
time out [of the office] going to the Gov-
ernor's office, briefings, and things like
that. I was essentially trying to keep the
environment thing going, but one thing
I would like to find a way to avoid in the
future is having people snapped away
like that. (489)
UNCERTAINTY OVER STATE NEEDS
Another factor affecting communications be
tween the State and the utility was uncertainty
over what information TMI was to transmit.
Dubiel said that he tried to convey the general
tone he picked up from the operations people, in
addition to relaying radiation information. (490)
He noted, however, that it was unclear what in-
formation was to be provided and that the plant
conditions to be conveyed in the course of making
offsite notifications were not clearly delineated in
the Met Ed Emergency Plan or its implementing
procedures. (491)
Indeed, the plan focused primarily on measure-
ments of radioactivity onsite and offsite and dealt
very little with plant status. (492) Transmission
of information about plant conditions had been
largely restricted to the operability of certain
plant systems at the time of initial notification. 140
(493) Further, the plan focused on current levels
of radiation, and not on the potential for worsen-
ing ones, given changes in plant status. (494) As
Dubiel explained :
. . . There is not explicit guidance to
state that if one believes the conditions
are going to get significantly worse, or
whatever, that additional or more con-
servative protective actions may be taken.
(495)
Throughout the day, the TMI personnel relay-
ing information to 'BRP were primarily non-
operations personnel. 141 Sometimes Dubiel, in
charge of radiological protection for TMI, would
talk over the Unit 2 phone line, but with the excep-
tion of Millers conversation with Dornsife, (497)
Unit 2 operations people were generally not talk-
ing directly to the State. Furthermore, us was the
case with the NRC, most of the communication
was from the Unit 1 control room. 142 (498)
Dubiel described his contact with the State as
periodic but not systematic: (499)
Well, I was periodically talking to them.
I don't think there was anything that you
would call systematic, ... I tried to pre-
sent the status of the plant to the degree
that I could understand it, to the degree
that any of us could understand. (500)
Dornsife told Special Investigation staff that
early on discontinuity in terms of the TMI people
with whom he spoke compromised the accuracy
of the information he was receiving:
... I was aware that up until I was
briefed by Gary Miller [around 9 :30 a.m.]
that I was getting somewhat disjointed
information ... I was aware I wasn't get-
ting real accurate continuous informa-
tion. (501)
Finally, Dornsife noted that at times it was
difficult 'for BRP to get through to TMI. BRP
personnel continuously monitored their end of the
open line over a speaker phone, but TMI employ-
ees only picked up their end every 15 or 30 min-
utes. (502) Sometimes BRP would have to get
their attention by shouting into the phone or by
calling an outside line and asking a Met Ed em-
ployee to pick up the open line. (503)
133 In addition, Dornsife had worked more than six months in the Burns and Roe home office on the TMI-2 design,
and then spent more than six months onsite in 1976. During the licensing process for TMI-2, all of Dornsife's time was
spent reviewing TMI-related licensing documents. (486)
140 For Miller's assessment, see Addendum 19, p. l.">9.
"' See also Addendum 20, p. 159.
'"After the evacuation of Unit 2, nearly all communications were with non-operations staff in 1'nit 1. (4!K>)
136
QUALITY OF THE DATA
Despite these problems. Dornsife did not ques-
tion the veracity of the information lie was receiv-
ing:
We think the utility was being perfectly
candid with us. We were asking questions
and they were giving us what we still feel
was accurate information. (504)
Hut the utility's ability to respond was limited
by its confusion over plant conditions, the rapidly
changing parameters and the state of crisis. Gary
Miller recalled two weeks later:
One thing . . . that stands out clear in my
mind, is that as the emergency director
and station manager during that day. I
was consistently pulled to the phone by
senior persons in the State_Government.
the XRC. and my own management, both
here and in remote locations. This caused
the pressure to be intense, as it was very
hard to concentrate on what I considered
to be a very serious situation ... I felt
strong in my obligation to the public and
to making sure that there was no [radia-
tion] emissions and that there was evacu-
ation in plenty of time if that was
i-equired. But the phone, the pressure, the
fact that the plant was in a state that I
had never been schooled in combined to
make it almost intolerable. (505)
NRC'S INTERNAL COMMUNICATIONS
Meanwhile, the XRC was still having difficulty
with its internal communications, difficulties that
did not diminish until late in the afternoon. Be-
cause of delays in getting briefed on what had
occurred, in setting up the utility's Emergency
Control Station in Unit 1, and a shortage of res-
pirators in Iwth units, an hour-and-a-half had
passed before two members of the onsite inspec-
tion team could enter the control room at Unit 2.
where the accident was being managed. (506)
Once they arrived in Unit 2. communications
were reestablished between Region I and the Unit
2 control room, with the XRC inspectors man-
ning the link. This link would be XRC head-
quarters" only one with Unit 2 until 4:30. and.
again, it was indirect. At 12:30 it finally estab-
lished a direct, o-way line that incorporated both
the region and the site, but the link-up at TMI
wa- to Unit 1. (507)
Both these communications channels were to
prove unsatisfactory. For example, for the first
time, shortly after 12 noon, the XRC asked that
the regional office inquire about incore thermo-
couple temperatures. 143 (508) Three-and-a-half
hours passed before there was any follow-up on
this question. The XRC headquarters was unable
to get incore thermocouple readings, the direct
indicator of core temperatures, and continued to
receive inaccurate hotleg temperatures, the indi-
rect measure. 144 (509)
The XRC's problems in communicating with
the plant through Unit 1 are illustrated by an ex-
change at 12:27:
REGION 1 : What are some of their [the
operators] ideas that they are using, you
know, relative to handling this situation,
you know ?
INSPECTOR ix UXIT 1 CONTROL ROOM : I
have no idea because we are in the other
control room, you know, that's all going
on [in] the other unit's control room.
(510)
Erroneous information was transmitted even
after the three-way communications link was set-
up. In an early conversation over this line, Chick
Gallina, one of the XRC onsite inspectors, re-
ported the following from Unit 1 :
GALLINA : Okay, the reactor pressure is
500 [psi].
IRACT : Got it.
GALLINA: Temperature 250 [F].
Okay?
IRACT : Got it, thank you. (511)
The information was in error. Temperature in
the primary system at 1:30 p.m. was not 250F:
only the coldleg approximated this temperature.
The hotleg was hundreds of degrees higher.
This incorrect information was quickly trans-
mitted to the Commissioners. At their 1 :> brief-
ing. Case and Davis, on the basis of the last er-
roneous hotleg temperature readings, told Com-
missioner Gilmsky and several members of the
Commission staff the following :
CASE: . . . system pressure has gone
down from 2000 [pounds] to ... 500.
Temperature is 250 degrees. Shortly they
ought to be going on the [cold] shutdown
decay heat removal system which can be
activated at 450 pounds [sic] ... So they
are reaching the point where things will
get stable in the primary system.
GILINSKY: Okay. So how do you feel
about the fate of the reactor ?
CASE: I feel good. Xow I get the im-
pression that it ? s stabilized or directly
approaching a stabilized situation. (512)
10 See Addendum 21. p. 150. for the text of the conversation.
"* See Addendum 22. pp. 159-160, for an example of the inaccurate transmission of hotleg temperatures.
137
.-;
- - -
NRC STAFF: . . . the hotleg, coldleg
situation, is there anything new on that?
DAVIS: As I understand they're both
reading 250 degrees. (513)
Throughout the early afternoon, TRACT was
receiving its information from, and relaying
questions almost exclusively through, NRC in-
spectors in the Unit 1 control room or in the re-
gional office. 145 Region I was having difficulty
acquiring accurate information from the onsito
team and transmitting it to headquarters. And
neither was responding effectively to State needs,
as reflected by communications among the State
agencies, or to the public.
A DECISION TO DEPRESSURIZE AGAIN
Sometime after 1 :10, the emergency command
team again decided to depressurize ; on this occa-
sion, the intent was not to assure the core was
covered, but rather to reach the point at which
the low pressure decay heat removal system could
be used.
This objective was not widely known within
the control room at the time. Miller, Ross and
Rogers, in a meeting two weeks after the acci-
dent, discussed their perceptions of the motive
behind the second depressurization :
MILLER: "VVe were kidding ourselves.
We were hoping for decay heat, you know
it?
Ross : That's where we were going . . .
MILLER: That's [expletive deleted]
[un] believable !
Ross : . . . I was just making a run for
decay heat.
ROGERS : No. No. We didn't want to go
to decay heat, not with those legs full of
steam.
MILLER : Ross and I did.
Ross : We talked about it.
MILLKI: : We say that as our . . .
Ross : . . . our only hope. (515)
THE HYDROGEN BURN l46
At 1 :50 p.m., Zewe directed the operators at
the front panels in the control room to open the
block valve to begin depressurization. As it was
opened, pressure in the containment shot up dra-
matically, reaching 28-31 psi, according to a strip
chart. 147 Since the start of the accident, contain-
ment pressure had never been greater than about
4 psi. (516)
Mehler was in the shift supervisor's office at
the time. He looked out into the control room:
What I noticed [was] the people started
to move a little faster, they were securing
pumps. So essentially, I thought we had
an ES again, which is an emergency
safeguard [actuation], but I didn't know
whether it was [due to] low pressure [in
the primary system] or reactor building
[containment] pressure. I have never
seen reactor building pressure go that
high. We went out to see what was going
on. (517)
The Spray Pumps Come On
Mehler left the office and went over to the con-
trol panels where Zewe, Shift Supervisor Joe
Chwastyk, Mike Ross and the operators were
standing. There he saw something he said he had
never seen before. (518) Indicators showed the
containment spray pumps were running. He
explained:
... To start spray pumps [in the contain-
ment building] you need 30 pounds of
pressure
and thev were running. I
couldn't believe that. I looked at them
[the spray pump indicators]. I walked
over and looked at the [pressure strip]
charts and that's when I saw the line
straight up and straight down. It looked
like somebody played with the transmit-
ter. It couldn't have been that [because]
we wouldn't have gotten the spray pumps.
(519)
The spray pumps never before had come on at
Three Mile Island, and by the afternoon the news
had spread to Unit 1. (520)
Mehler told Special Investigation staff that
when he was at the control panel, he was standing
beside an NRC inspector, whom subsequently he
said he could not identify. 148 The inspector asked
why the pumps were running. Mehler explained
that they were designed to lower containment
pressure and were activated at 30 psi. (522) He
pointed out the spike on the pressure chart. Ac-
cording to Mehler :
. . . He [the inspector] asked me why I
was concerned because the spray pumps
145 IRACT personnel were aware, at the time, of the limitations in the communications system. (."14)
141 Because of conflicts in the recollections of those present, it is not possible to reconstruct a consistent account of
what occurred in the control room at 1 :50 p.m. In general, evidence adduced liy the Special Investigation supports the
findings of the XRC's Special Inquiry Group, which explored the issue in far greater depth than any of the other inves-
tigations. What follows is a synthesis and summary of a large body of contradictory evidence.
141 The chart is difficult to read at a glance and has been interpreted variously as having read 28-31 psi. The con-
tainment is designed to withstand a pressure of 60 psi.
148 Mehler attempted to identify the inspector for the NRC Special Inquiry Group. His description did not fit any
of the inspectors in the control room. (521)
138
were running. I told him they would only
start at 30 pounds. I walked over to the
chart ; 31, it was straight up. I looked at
it and said, "That's impossible." I showed
it to him. He didn't know what was going
on. All he did was write down what we
told him . . . Then he went back in the
office after we secured from all that.
(528)
The Pressure Spike
As noted, there had in fact been a sharp in-
crease in containment pressure a so-called con-
tainment pressure "spike," and pressure had gone
to 28-31 psi. With the containment spray pumps
running, containment pressure decreased rapidly,
and in about five minutes the spray system was
shut off.
Most of those who were aware of the pressure
spike attributed it to an electrical malfunction
in the instrument. (524)
Zewe said :
. . . My first reaction was I stepped back
and looked at it [the pressure spike] and
said "What in the world was that ?" to
all that were there ... I conversed with
the other shift supervisor there and also
Mr. Ross, who was there, and we con-
cluded that it was just some phenomenon.
some voltage spike or transfer that af-
fected the recorder or pressure indication.
The spike had resulted from the rapid com-
bustion of hydrogen within the containment. The
hydrogen had been produced during the earlier
period of core uncovering, when core temperatures
were in excess of 2.500 F. 149 Hydrogen was re-
leased to the containment when the block valve
was opened to vent or control pressure.
A Strange Noise
There were other symptoms of the burn. Miller,
Logan. Dubiel. Roge'rs. Flint and a Met Ed en-
gineer named Walter "Bubba" Marshall also were
in the control room, but not at the front panels. At
the time of the pressure spike, they heard a noise
that was inaudible to the people" at the control
panels. (527)
Whether the noise was caused by the hydrogen
burn has been the subject of considerable post-
accident analysis. Miller and Dubiel stated they
commented on it at the time :
MILLER: ... I was aware of a noise
. . . and in fact. I believe I asked, "What
was that?" in fairlv strong language.
(528)
DUBIEL: ... I said, "It sounded like
the ventilation system." (529)
According to Ross, Miller then spoke with him:
. . . [Gary] said, "Did you feel that or
hear that," something to that effect. I said
"no." In fact, I remarked to him, "This
is not the time to get nervous ..." I spec-
ulated the noise he heard could have been
the ventilation. He seemed to think it was
right above him in the duct work. (530)
To Miller, that explanation seemed reasonable :
... I did not closely evaluate [it] be-
cause I was told, I believe, that it was a
ventilation system which was changing
modes and did make a thud-type noise
[when that occurred]. (531)
Others who heard the noise also concluded it
was the ventilation system. (532)
The Special Investigation staff found that the
noise probably was made by the ventilation system.
There are butterfly valves directly over the control
room which could have been thrown shut when the
emergency safety system was actuated because of
the high pressure in the containment building:
Question : Well, if the ventilation sys-
tem were on at that point in time, and
ESFAS [emergency safety systems]
came on, wouldn't that close the damp-
ers?
Ross : That would put it on recirc [illa-
tion], yes.
Question : Might that have been what
he heard ?
Ross: It's possible, very possible. I
hadn't even thought about it, but it's very
possible. (533)
After hearing the noise and concluding that it
was the ventilation system. Flint glanced over
the control panels, where he learned that the opera-
tors were concerned about a pressure spike. He
noted that they were checking "the possibility of
an electrical problem." (534)
Dubiel also had moved closer, over to the far
left panel on the console. He overheard the opera-
tor in front of the spray pump controls indicate
that they had come on. A short while later he
looked at the strip chart and noticed the spike.
(535)
Ross, after speaking with Miller, looked over at
the control panels. He found the same things as
the others. Since the pressure immediately went
back to zero, he said, "We didn't try to analyze or
deduce. We wrote if off.'' (536)
See p. 108 for a description of the production of the hydrogen. The 2.300 reading is known from the incore thermo-
couples (see pp. 113-114). However, core temperatures were certainly much higher. The President's Commission esti-
mated temperatures in the core to have reached 4,350-1,500 F. (526)
139
The Symptoms Are Not Understood
Rogers said that "just about everyone in the
control area heard the, noise." (537) Zewe said he
could not understand how anyone could have over-
looked the other indicators the pressure spike and
the activation of the spray pumps :
... I cannot honestly see how . . .
anyone that was there that had any con-
cern at all could overlook [those indica-
tions] because we certainly stopped every-
thing, and that was the main thing that
was in progress at that point in time.
(538)
Yet Zewe, Ross, Mehler, and operators at the
console said they heard no noise, and Rogers,
Miller and Logan stated they were not aware of
either the pressure spike or the spray pumps. (539)
George Kunder and the two NRC inspectors in the
control room said they were unaware of any of
these phenomena. (540) The evidence suggests that
Flint, Marshall and Dubiel were the only ones in
the control room to have heard the noise and who
also were aware of both the spike and activation
of the spray pumps. (541)
Only Chwastyk and Mehler have said they
recognized that the spike reflected a real increase
in containment pressure. Chwastyk was the only
one who said he concluded it was the result of a
"hydrogen explosion." (542)
There is conflicting evidence as to whether Miller
was made aware of the pressure spike or the actua-
tion of the sprays before he left for a briefing of
the Lt. Governor at around 1 :55 p.m. He has con-
sistently maintained that he was unaware of either
event.
The contradictory evidence stems in large part
from testimony and statements made by
Chwastyk and Mehler that they believed Miller
was informed of the hydrogen burn or related
phenomena. Chwastyk testified before the NRC
Special Inquiry Group (SIG) that his "best
recollection" was that he told Miller they had
experienced "some sort of explosion." (543) He
stated that he believed his conversation with
Miller occurred in the context of discussing an
attempt to reestablish a bubble in the pres-
surizer.
Miller and Ross both testified before the SIG
that they were unaware of any such attempt,
which appears to have led the SIG to doubt
whether Chwastyk actually mentioned the burn
to Miller that afternoon. 150 (549)
Similarly, although Mehler and Chwastyk re-
called discussing the event with an NRC inspector,
neither of the two NRC inspectors in the Unit 2
control room at the time recalled being aware of
the pressure spike, actuation of the pumps or any
of the other phenomena related to the hydrogen
burn. (550) One. James Higgins. who indicated
he would have been the more likely of the two to
look at the panel with the strip chart, said his
first knowledge of the spike came on Friday morn-
ing. March 30. He noted that at the time he might
have looked at the strip chart, visible through a
window approximately four inches wide, shortly
after the spike was recorded :
. . . and the spike would have been there
and I would not have considered it sig-
nificant. I may have just looked at where
the reading was at that time, knowing
that it had been 2 pounds the last time
I looked at it and it is now reading 1 [to]
3 pounds ... It was always the type of
thing where I had a backlog of about 40
questions I was supposed to answer for
the people, in Washington. (551)
There is evidence suggesting that Higgins did
look at the chart. A Region I "Incident Message-
form" shows that at 3 :45 p.m.. Higgins reported
containment pressure of psi to the NRC, five
minutes before the segment of the strip chart
showing the spike would have, disappeared from
view. (552) In his report to the Region he made
""Rogers told Special Investigation staff that when he was looking at the steam tables (presumably in the after-
noon ) , it was in relation to an attempt to redraw a bubble in the pressurizer, lending credence to Chwastyk's recollec-
tion of the timing of his discussion with Miller. (544)
Mehler told the SIG that he thought he had informed those who were in the shift supervisor's office of the spike,
and that Ross and Miller were among those present. (545) However, Ross told Special Investigation staff that he was
out in the control room at the time of the spike, where he discussed the thump with Miller. (546)
The matter is further complicated by other contradictory evidence. See, especially, NRC Special Inquiry Group, Vol.
II, Part 3, pp. 138-152. The SIG concluded :
Based on the weight of the evidence, it appears more probable that if Miller learned of the reactor building
pressure increase, it was in the context of an indication that was not understood or was discounted as an elec-
trical malfunction, rather than as a possible hydrogen explosion. If Miller was in fact informed of the pressure
increase or was aware of it at 1 :50 p.m. on March 28, it is impossible to determine from available testimony
whether it is most likely that he subsequently forgot the event, or if he simply failed to take account of what
was happening, or if he has testified falsely about not recalling learning of it at the time. (547)
On March 21, 1980, the NRC Commissioners directed the Office of Inspection & Enforcement to review the transfer
of information between the utility and the NRC to determine whether a further civil penalty to Met Ed is justified.
(548)
140
no reference to a spike, even though it would have
been visible at the time. 151
The Hydrogen Burn Is Real
Days later, when the control room strip charts
were analyzed, the utility concluded that there
were too many redundant indications from the
control room instrumentation for the hydrogen
burn to have been anything but real. (554)
The containment building had automatically
isolated again, the containment sprays had come
on. emergency core cooling was initiated auto-
matically at full flow, and the wide-range pressure
recorder, which is tied to containment pressure,
had a small spike on it. (555)
Each of these indications appeared on instru-
ments in the control room. The evidence suggests
that confusion in the control room over the source
of the thump and over persistent electrical prob-
lems around that time diverted attention away
from those indicators. (556)
There is no direct evidence of any deliberate
effort by utility personnel to conceal from the
XRO. the State or the public information on the
hydrogen burn or on uncovering of and damage
to the core.
Rather, failure to recognize the hydrogen burn
and its meaning can be partially explained in the
context of several other factors. One was a dis-
ruption in the management of emergency opera-
tions at the plant when, around 1 :55. Miller and
Kunder left to go brief Lt. Governor Scranton in
Harrisburg. 1 " 2 Another involved the incore ther-
mocouple readings. As noted. Miller and Porter
had discounted them as unreliable. At the, higher
temperatures indicated by the thermocouples, fuel
failure was inevitable, as was the generation of
hydrogen. Instead, the hottest temperatures of
which Zewe. for one. was aware were the hotleg
readings in the neighborhood of 800F. He and
some of the other control room personnel said,
therefore, they did not suspect that the threshold
temperature for a zirc- water reaction (1,600F)
had been passed. (557)
ARNOLD QUESTIONS CORE STATUS
At approximately 2 p.m., 10 minutes after the
spike. Rogers and one of the plant managers were
in the shift supervisor's office talking by telephone
with (JPF Vice President Robert Arnold. The
conversation centered on his concern whether
the core was covered. Arnold recalled speaking
with Rogers and someone else, whom he believed
to have been Logan. (558) This telephone call
might explain why Rogers and Logan said they
did not learn of the pressure spike or activation of
the spray pumps during the first day.
Arnold was assured by the two the core was
covered on the basis of the experience with the
flood tanks. As he recalled :
. . . They felt at the time that they [had]
sort of passed the crisis, as it were, and
the core flood tanks were indicating that
the core was covered, the system was full
[of water]. (559)
Arnold said he questioned their conclusions :
... I believe I indicated to [them] at that
time my uneasiness as to whether that, in
fact, was that reliable an indication and
told them I thought they ought to review
very carefully whether or not in fact they
had the core uncovered . . . certainly my
impression was both a plant staff member
and Lee Rogers were confident the core
was covered. My recollection was the com-
ment was made they didn't think it had
ever been uncovered. (560)
DEPRESSURIZATION FAILS
Ross and Miller's attempt to bring reactor pres-
sure down to the i>oint where the decay heat re-
moval system could be brought on lasted a little
over an hour. At 2:34 p.m., pressure fell to 410
psi, 25 psi below the lowest pressure achieved
previously. The core flood tanks again injected
water onto the core. It is estimated that this in-
jection lasted two minutes and added another 165
gallons of water to the primary system. Pressure,
however, not only stopped falling, it rose 10 to 15
psi, leveling off at 420-425 psi. It remained at that
level until operators closed the block valve at 3 :08
p.m., terminating the attempt. (561) The inability
of the utility to bring pressure lower was not re-
ported to the NRC for several hours. 183
Why Pressure Stabilized
One of the continuing mysteries of the accident
is why pressure could not be lowered further, even
with the block valve open. A number of plausible
theories has been advanced.
One is that superheated steam was being pro-
duced while the core was uncovered, tending to
keep pressure up. 154 (562)
" There were a number of electrical instruments that malfunctioned at about this time because of a loss of power
in two electrical busses.
In explaining to the Special Investigation how he might have looked at the four-inch-wide pressure strip chart
without noticing where the needle had gone before, Higgins referred to spikes on other monitors caused by electrical
malfunctions in plant instrumentation. (553)
' M This is discussed further on p. 144.
ia See p. 143.
"" See p. 107 for an explanation of the phenomenon.
141
Another, held by analysts at the Nuclear Safety-
Analysis Center (NSAC), 155 is that the amount
of cold water being added to the core through HPI
was sufficient to balance the steaming in the hot-
legs, causing pressure to stabilize. (563)
The primary difference between these two theo-
ries is where they say the interaction of steam and
water occurred in the core or above it.
As on previous occasions, control room personnel
had differing recollections about the purpose of
depressurization and why pressure would not go
lower. Rogers said he was not aware that pressure
could not be taken lower because he was not aware
that Miller and Ross were attempting to bring on
the decay heat removal system. 150 Rogers told the
Special Investigation staff :
... I am saying that [the fact that pres-
sure could not be taken lower] was not
information that was readily known at
the time [because it was not readily
known] that we were going to go any
lower. (564)
Zewe, on the other hand, attributed the in-
ability to lower the pressure to saturated condi-
tions in the primary system :
... as I recall, we were unable to get be-
low 410 to 420 pounds . . . We kind of de-
duced [that was] because of saturation
pressure in the cooling system at the time.
(565)
Ross, Zewe's supervisor, had no recollection of
analyzing the difficulty in those terms :
I was aware we were hot; I don't think
I was aware that we were actually super-
heated in the steam. I don't think I ever
deduced anything about superheated
steam. (566)
Ross could not recall any analysis that after-
noon of why pressure could not be brought lower
during depressurization :
... I don't think we ever said, "Why
won't the pressure go any lower?" I don't
think we ever sat down and said, "Why
won't it go any lower?" I don't think we
ever analyzed that. (567)
HOTLEG TEMPERATURES
Another related mystery concerns the measure-
ments of temperatures in the hotlegs during this
period. These temperatures had fallen dramati-
cally after the hydrogen burn. Those in the "A"
loop dropped sharply from about 715F at 1 :45
p.m. to about 460F by 3:15 p.m. When they
reached 620F, they came back onscale on the
control room console monitor. They also fell be-
tween 3 p.m. and 3 :15 p.m. to the point where
superheated steam was no longer indicated in the
hotleg. However, after 3 :15, following closure of
the block valve, they increased to the point where
superheated conditions were again indicated.
They remained there until about 5 p.m. 157 Still
unresolved is whether this indicated a second un-
covering of the core or is attributable to other
factors.
In the judgment of Special Investigation staff,
neither the President's Commission nor the NRC
Special Inquiry Group has fully explained this
phenomenon.
Analysts at Battelle Columbus Laboratories,
who performed the analysis for the NRC
Special Inquiry Group, postulated that the re-
turn to superheated conditions resulted when
the hot piping in the system heated the steam
and gas in the hotlegs to that point. (568)
According to analysts at NSAC, the tempera-
ture fluctuations can be explained by the heat-
ing effect of fission products plated along and
throughout the primary system fission products
that were distributed throughout the system, in-
cluding the hotlegs, following uncovering of and
damage to the nuclear core early in the morn-
ing. 158 (569) The effect of such plating would
be to provide a source of heat for the produc-
tion of superheated steam throughout the sys-
tem, and not just in the core. This plating
155 See "Glossary of Organizations," Appendix F, p. 381.
""See p. 138.
57 The correlation between the hydrogen burn and the simultaneous temporary unblocking of the hotlegs has not
been explained.
An analysis of the plant data shows that the hotleg temperatures began to converge with the temperatures of the
coolant in the surge line (the pipe running from the hotlegs to the pressurizer) following the hydrogen burn. By
3:08 p.m., when the second attempt at depressurization was concluded, both the surge line and the hotleg temperatures
were at the boiling, or saturation, point. Thereafter, the hotleg again showed superheated steam conditions, while the
surge line remained superheated or at the boiling point until a later decision to repressurize.
The hydrogen generated during core uncovery early in the accident is assumed to have accumulated in the hotlegs
mid to have mixed with the superheated steam there, helping to block the flow into the steam generator and contributing
to the stagnant, superheated temperatures in the hotlegs. Special Investigation staff theorize that when the pressure
spike occurred at 1 :oO p.m., after the opening of the block valve, hydrogen gas in the hotleg may have been vented out
through the pressurizer, allowing flow to return through the hotleg and causing the temperatures to fall. Readings once
again appeared on the resistance temperature detector, the hotleg temperature measuring device, which for awhile
may have reflected temperatures of coolant flowing through the core.
158 See p. 124 for further details on plating.
142
could have further heated the water and steam
in the hotlegs to superheated temperatures.
Analysts from Battelle Columbus Laboratories
tincl this theory to be implausible. (570)
A POSSIBLE SECOND UNCOVERING
It is also possible that the superheated tem-
peratures reflected a second uncovering of the
core. That could explain why the hotleg was
filled first with saturated steam and then again
with superheated steam.
In analyzing the question of a second core un-
covering, the staff of the Special Investigation at-
tempted to calculate the rate at which coolant
was injected onto the core.
The second depressurizmtion took place from
!:!.> p.m. to 3:08 p.m. Using the NRC figure
of an average flow rate of 150 gpm for the
entire period from 1 :15 p.m. to 5 :20 p.m., 159 along
with other data, the staff estimated that the
average net flow rate for the first two hours was
about 100 gpm of water. (571) This is only 30
gpm greater than the net average flow during
the morning hours when the core is known to
have been uncovered.
There are. however, several consideration?
about conditions during this time. First, decay
heat was lower during the afternoon. Second,
the amount of water then being released through
the let-down system is not accurately known.
Finally, in general it is very difficult to estimate
and compare flow rates at various times based
on the available data. Thus, it is hard to use
the estimated rate of flow to determine whether
the core was uncovered.
There is insufficient evidence for the Special
Investigation staff to conclude which, if any.
of these theories is correct. 160 Such a determina-
tion may be possible when the core can be
examined directly.
WHY HPI WAS THROTTLED
While the calculations of the XRC and Special
Investigation staff provide only estimates of ac-
tual flow rates, they still raise the question of why
utility personnel throttled the amount of water
delivered to the core to such an extent.
After the first depressurization. a number of
utility personnel had concluded (based on the
minimal injection of water from the core flood
tanks) that the core never had been uncovered.
Thus the control room personnel believed they
could use the make-up pumps to cool the core, even
at low rates of flow :
Ross : We felt that we had the core cov-
ered; we felt that we were cooling the
core with the High Pressure Injection
which we maintained throughout this
time. (573)
MILLER: That day. I don't feel from
7 in the morning on[.] that we felt we
had uncovery or maintained uncovery. I
don't think "we had the time to think
about the hours before that and what they
might have done to the condition of the
core. We knew they damaged it, and we
knew the systems we had [High Pressure
Injection] were the only systems we had.
and they were working effectively. 1 ' 1
(575)
RIGHT HOTLEG TEMPERATURES . . .
The transcripts of the NRC tapes show that
TRACT received another report on hotleg tem-
peratures at -2 :20 p.m. According to the evidence,
this was the first accurate one received since the
beginning of the accident. The temperature was
said to be 600F, reflecting the return to onscale
readings on the control room console. 1 * 2
Even that lower temperature, when viewed in
conjunction with primary system pressure, indi-
cated superheated conditions.
Thirty minutes later a hotleg temperature of
550F was reported: it also reflected superheated
conditions. This report was to be the last received
by TRACT over the next several hours.
. . . BUT OTHER MISINFORMATION
Although the XRC was at last getting accurate
hotleg temperature readings, it still was not get-
ting accurate information on natural circulation.
At about 3:15 p.m., TMI Unit 1 Shift Supervisor
Greg Hitz informed TRACT that the plant was
being cooled with natural circulation at a rate of
3F per hour. (576) In fact, for the next four-
and-a-half hours, there was little or no heat trans-
fer by natural circulation through the steam gen-
erators. 163 (577) More important, there was a pe-
See Addendum 23. p. 160. for the NRC's calculations.
* There was also an unexplained slight upward trend in the source range neutron monitors during the afternoon
hours. However, the monitors do not appear to have liehaved as they did in the morning when there is no doubt the core
was uncovered. If the core were uncovered again during this period, it was probably a result of depressuriziug without
providing sufficient high pressure injection to the core. (.TTiJi
m At this writing. XSAC is in the process of preparing a report, to include estimated high pressure injection flow
rates during the accident. t ."4 i
*" See p. 142. text and accompanying footnote.
There was very minimal natural circulation, not enough to state it was successfully established.
143
riod after 3 :08 p.m., when the core was not being
cooled at all by natural circulation. As noted, the
hotleg readings indicated the core may have been
uncovered again : temperatures in the hotlegs re-
turned to stagnant, superheated conditions, and
there was no flow through the primary system.
(578)
COMMAND TEAM FRAGMENTED
As described earlier, during the second attempt
at depressurization, Miller, the Station Manager
and Emergency Director, and Kunder, Superin-
tendent of Technical Support, left the plant and
joined Jack Herbein, Met Ed Vice President for
Generation, to go to Harrisburg to meet Lt. Gov-
ernor Scranton. (579) As part of Scranton's ef-
forts to understand what was happening at the
plant, he had requested that Walter Creitz, Presi-
dent of Met Ed, provide an authoritative report
from someone with firsthand knowledge of plant
conditions. (580) Scranton's office had not asked
for any particular individual, and it is unclear
who decided that Miller was the appropriate per-
son, despite his role at the plant. Herbein said:
I felt it was appropriate to take any
member of the plant staff Avith me for
response to any detailed questions re-
garding plant status that might arise in
our session with the Lt. Governor. 184
(583)
BRIEFING STATE OFFICIALS
According to Paul Critchlow, the Governor's
Press Secretary and Communications Director for
the State, the meeting was strained because it ap-
peared that Herbein was not planning to tell the
State, of radiation releases that had occurred
earlier that day. (584) At a press coTiference at the
TMI Observation Center prior to leaving for the
State Capitol, Herbein had not mentioned them.
(585) Critchlow said that State officials were very
concerned, as they believed they should have been
notified so that they could take whatever precau-
tions were necessary. (586) As it was, they had
received the information from the Bureau of
Radiation Protection, which had detected the
radiation.
At the briefing, Herbein was confronted on this
issue. Critchlow described the situation :
[Herbein] was asked, "Why didn't you
tell the press ?" He said he had never been
asked, or the question did not come up,
or something like that. That immediately
led to a very quickly developing caution
on our part in dealing with Metropolitan
Edison. (587)
Miller was noticeably upset and said very little
at the briefing, according to Mark Knouse, Scran-
ton's Executive Assistant. (588) Miller remem-
bered spending much of the time in the Lt. Gov-
ernor's office on the telephone, talking to the Unit
2 control room where he had left Logan in charge :
Most of the briefing was done by Jack
[Herbein]. I was there initially . . . and
for the most of that meeting I believe I
was on the phone to the plant ... I was
probably missing from half of that meet-
ing. (589)
Dornsife had not been invited to the 2 :30 p.m.
briefing and did not learn of it until it was in prog-
ress, (590) even though he was the only State of-
ficial equipped to deal with the technical informa-
tion being provided by the utility's operations
staff. His absence is even more noteworthy because
Dornsife accompanied Scranton to a television in-
terview at 2 p.m. in the Capitol building. He
then returned to his office across the street. (591)
DISILLUSIONMENT WITH MET ED
At 4 :30 p.m. Scranton conducted his second press
conference. He wanted to place the population on
alert without alarming them. (592) He made it
clear at the press conference that he had become
disillusioned with Met Ed and was suspicious and
mistrustful of the utility. (593)
Until then. Met Ed had been Scranton's primary
source of information. Having lost confidence in
the utility. Scranton and his staff sought another
source of reliable information. They turned to
Gallina and James Higgins of XRC Region I, who
would later provide briefings for the Lt. Governor.
The first such briefing occurred at 8 p.m. Xat
Goldhaber, Lt. Governor Scranton's Administra-
tive Assistant, indicated that the State found the
two NRC officials to be a great improvement :
... we felt that we were getting more ac-
curate information, more complete infor-
mation, and more technically qualified in-
formation than we had been getting ear-
lier during the day . . . The presence of
those specialists from a governmental
agency lent a certain feeling of confidence
in the reliability of the data that they
were providing. (594)
101 Herbein's recollection differed from those of both Miller and Kunder. Miller told the XRC that Herbein had di-
rected him to leave the plant for the briefing and that he had expressed his concern about departing. Kunder also re-
called that Herbein had "wanted Gary to go along and Gary said he wanted me to go along so I could back him up
with any answers to technical questions." (581) Herbein told the Special Investigation, however, that he "asked Gary
to release George Kunder" and that Miller "felt [that] if George was going to go, then he ought to accompany me also."
(582)
144
THE NRC AND PLANT CONDITIONS
By 4 p.m., at least one senior XRC official
Victor Stello believed the information received
by IRACT in the afternoon indicated the core
might be uncovered. (595) Since around 1 p.m..
IRACT had been receiving hotleg temperatures
and primary system pressure readings that, if true,
indicated to Stello that the reactor core was uncov-
ered. (596) He was still waiting for the incorc
thermocouple readings he had requested before
noon to verify the hotleg temperatures and confirm
or invalidate the indications of superheated steam
in the hotlegs and whether or not the core was un-
covered. (597)
INGORE TEMPERATURES REQUESTED
At 4 p.m. IRACT had still not received any
word on the incore thermocouples. Via the three-
way IRACT-Unit 1-Region I telephone line, Mike
WUber, the IRACT Field Communicator, at Stel-
lo's request, raised the issue again with the regional
office. Donald Caphton was manning the phone
there :
WILBER : Some time ago we asked about
the incore thermocouples . . .
CAPHTOX : Xo. I have no information.
(598)
Shortly after this conversation, Caphton had an
XRC inspector in the Unit 1 control room ask TMI
Unit 1 Shift Supervisor Greg Hitz to come to the
phone. IRACT asked Hitz to get the incore ther-
mocouple readings. (599)
THE QUESTION OF SUPERHEAT
While Hitz was still on the line, Stello asked to
speak with him. He raised the issue of the various
readings and their implications:
STELLO: Let me bounce a question off
you. If you really have 550 degrees on that
hotleg. it's clear that you're getting some
superheat. If you're getting superheat,
there's a chance the core could be un-
covered. The only way you're going to get
rid of that problem is to find a way to get
more water in that vessel and get that core
level back up. Have you thought about
what problem you've got, if indeed you've
got 550 degrees on that hotleg at 450
psi?
Hrrz : Yeah, I see what you're say-
ing. Okav I . . . They ... do have the
BWST [Borated Water Storage Tank]
lined up and 175 inches indicated in the
pressurizer. which means that the core
would be covered. They've also got the
core flood tanks floating on that.
STELLO: But that doesn't necessarily
mean that they don't have a steam bubble
in there,
Hrrz : Oh, okay, you're talking about a
steam bubble in the core.
STELLO : Yeah, and if you have a steam
bubble in the core, you've got the top part
of the core which could be uncovered
superheating the stuff coming out of there,
and that's what's giving vou the reading.
(600)
Hitz said he would raise the issue with his
counterparts in the Unit 2 control room. (601)
Hitz also explained to the regional office and
IRACT that he had spoken to Mike Ross and that
Unit 2 control room personnel believed that mini-
mal injection by the core flood tanks meant the
core was covered. (602)
Ross recalled speaking with Greg Hitz during
the day over the telephone connecting the Unit 1
and Unit 2 control rooms. However, he stated that
he had had no conversations with Hitz or anyone
else in which he was told that the XRC wanted to
know whether the utility had considered super-
heated conditions in the reactor. (603) He said he
was certain that Hitz had never mentioned super-
heated conditions and that he would have remem-
bered it had Hitz done so. (604) He explained:
... If someone came in and said we were
superheated, "you ought to do something,"'
I think we would have moved in on it. It
wasn't total bedlam. (605)
More generally, both Ross and Gary Miller told
Special Investigation staff that the XRC never
recommended that day that the utility pursue a
particular course of action. (606)
Incore Readings Not Available
Several minutes later, Hitz spoke with Richard
Keimig at the regional office on the three-way line :
Hrrz : First of all, I can't get the incore
temperatures, okay ?
KEIMIG : You cannot get them ?
HITZ : They [the computer] print out
question marks . . .
KEIMIG : Okay, what's that mean ?
HITZ : That means that either the com-
puter point is messed up okay.
KEIMIG : Yes.
Hrrz: Or that the line you know,
the where you sense it, that line's bro-
ken or something's messed up with that
line . . . They're tiying all of them to
see if we can get any of them to print,
okay?
KELMIG : All right (607)
Hitz could not recall subsequently who gave
him that information. (608) It did not correspond
to, or indicate awareness of, the existence of data
145
from equipment set up earlier in the day to ac-
quire readings of incore temperatures directly
from the thermocouple leads in the cable room.
In fact, there is no evidence that anyone had used
that instrumentation since around 9 a.m.
HOTLEG TEMPERATURE ANALYZED
At 4 :14 p.m., still questioning the status of the
core, Stello called Eisenhut and asked him to con-
tact Babcock & Wilcox to try to get a better un-
derstanding of the hotleg temperature readings.
(609)
Eisenhut replied that B&W was on the tele-
phone at that moment but "said they don't have
enough information to straighten it out either."
(610) Stello then spoke with Thomas Novak at
NRR and asked him to consider alternative ways
of increasing flow through the core to eliminate a
steam bubble. (611)
By 4 :24 p.m., statements by EMT members con-
cerning the status of the reactor were no longer
as optimistic as they had been throughout the
morning and early afternoon. Gossick told Ed-
ward Fay of the NEC's Office of Congressional
Affairs, "we're still all right, but we still don't
have this core the way we want it ... we just
can't say that we're stabilized yet." (612)
Stello again spoke with Eisenhut, who said that
the question mark readings for the incore thermo-
couples were at that moment being raised with
Babcock & Wilcox. He also told Stello that the
reason B&W was not concluding that superheated
steam was present was that their readings on sys-
tem temperature and pressure were from indica-
tors in the pressurizer, rather than from the hotleg.
Stello stated his disagreement with the B&W read-
ings and gave Eisenhut the readings he had for
pressurizer temperature, primary system pressure
and hot and coldleg temperatures.
Eisenhut soon concurred with Stello's opinion :
EISENHUT: You got it man. That's it.
They've got a problem.
STELLO: You're above saturation and
the only way that's possible is with su-
perheat. (613)
Stello told Eisenhut to give Babcock & Wilcox
the "right numbers." (614)
NRC: WHAT ACTIONS TO TAKE?
From the EMT office adjacent to IRACT, Gos-
sick called Acting Chairman Gilinsky, who was at
Commission headquarters. Gossick began :
We've got a little update here I think
we need to give you . . . Let me get
John Davis and Vic Stello on here to
give you the situation with the core that
we've got. We've got I think a significant
development coming up here. 165 (619)
With Davis and Gossick on the line, Stello ex-
plained to Gilinsky his concern about superheated
steam. Gilinsky responded, ". . . you're saying that,
in fact, the core may not be covered." (620) He
asked who was in charge at the plant, but neither
Stello, Gossick nor Davis knew. (621) Gilinsky
then asked whether there was "anything we ought
to do about that beyond having talked'' with Hitz.
(622)
Stello replied :
The only thing I can think of doing is to
use our minds and understanding and
tell them what we think based on the facts
we hear, and they must make the judg-
ment. We cannot make the judgment here
because we're relying on information
that's from too many different channels.
I don't have enough information myself
to decide what I would do. I can only re-
act to the facts and raise questions for
them to consider. (623)
Gilinsky suggested that "the natural way to
handle it" was to speak to the NRC onsite inspec-
tion team leader and have him raise the issue with
the licensee "to make sure that our message gets
through." (624) He suggested they "talk to the
superintendent" and said, "I think we probably
ought to get some feedback." (625) Then the fol-
lowing exchange took place :
GOSSICK: We've got to be careful that,
you know, they don't start asking us what
to do and then . . .
GILINSKY: No. They're in charge, and
we can only offer something that we
thought of, but they are absolutely in
charge. There can't be any question about
"None of the three EMT members have recalled having learned on March 28, 1979, that there was superheated
steam in the primary system. In an interview with Special Investigation staff, Davis said he did not remember learning
of superheat, even though the tapes indicate he was a party to the conversation with Commissioner Gilinsky and Stello
concerning superheat. (615) Gossick recalled Stello's concern, but not that it arose on Wednesday, even though lie placed
the call to Gilinsky, put Stello on the line to brief the Acting Chairman on "a significant development' 1 having to do with
"the situation of the core," (616) and participated in the ensuing discussion about superheated steam and core uncovering.
Case testified on the subject before the Subcommittee. He originally said that he did not know about superheated
steam until late in the afternoon. He said he had told the Senators during a 5 :10 p.m. briefing on March 28 that "it is
not completely clear to us that even though the core is covered there might not be a steam bubble someplace in the
core," (617) because he had learned about the superheated steam. However, later in his testimony, he conceded that
there was a significant difference between superheated and saturated steam and stated that he was only aware that
afternoon that there was a steam bubble, not that it was superheated. (618)
146
that. And we don't want any confusion
in anybody's mind, especially in their
mind.
GOSSICK : That's right
GILIXSKY : And they've got to assess ev-
en-thing that, you know, that they need
to assess.
STELLO: We'll make it very clear to
them that the decisions that are being
made are theirs, and that the only thing
we're doing is asking questions. (626)
CONCERN ABOUT SUPERHEAT
While Stello. Gossick and Davis were speaking
with Gilinsky. IRACT established its first direct
communications channel with the Unit 2 control
room. (627) It was 4:36 p.m.. over 12 hours since
the accident began. When Stello returned to the
IRACT office from the EMT office, he asked
Moseley to raise the issue of superheated steam
with .fames Higgins. the XRC inspector in the
Unit 2 control room. While Moseley spoke with
Higgin? over the telephone, Stello stood next to
Moseley (Stello's voice was also recorded on the
tape) :"
STELLO (to Moseley) : Let's get some-
body to explain the 580 degree hotleg
temperature.
MOSELEY: The high hotleg tempera-
ture, [do] you conclude that there is su-
perheat there ?.
HIGGIXS : The hotleg?
MOSELEY : Yes.
HIGGIXS: There probably is. I'm not
sure.
MOSELEY : How do we know there's not
a steam bubble in the reactor itself and
what the level is in the reactor; is all the
fuel cool I
HIGGIXS : They're not positively certain
that there's not a bubble in the reactor
vessel . . . they're not 100 percent certain.
(628)
Higgins explained, as Hitz had. the command
team's interpretation of the partial injection by
the core flood tanks. (629) He also said they had
ruled out any attempt at rapid depressuriza-
tion 1S6 the step which Stello and Moseley had
believed would have to be taken. (631)
Stello. Moseley and Higgins continued :
STELLO : Does the licensee understand
- i degrees in the hotleg?
MOSELET : It means that it is superheat ;
they concede that,
STELLO : They agree to that I
MOSELET : Yeah.
STELLO: Do they have any way to ex-
plain superheat without the core being
uncovered 3
MOSELET: Not to my satisfaction, no.
STELLO : Did you ask ?
MOSELET (to "Higgins) : Have you pur-
sued with them this question you and I
talked about a little earlier, and that is,
how do we know that the core is not un-
covered, partially ?
HIGGIXS: We have talked that over.
Actually, most of the discussion on that
was between the people here on site
the unit superintendent Bob Arnold . . .
the vice president of Met Ed in a dialog
of about 20 minutes or so and I listened
to the whole discussion. The final result
of it was that they felt very confident
that the core was covered, based on indi-
cations when they were blowing down
and the core flood tanks and the interac-
tions there, although they could not real-
ly give assurance of 100 percent that the
core was covered.
MOSELET: Well, the core flood tank
story is not convincing to me. (632)
Moseley then turned the phone back over to the
field communicator.
At the height of XRC concern over uncovering
of the core. Stello and Moseley were on the phone
with an XRC inspector in the Unit 2 control room.
They learned that the utility agreed there was
superheated steam in the hotlegs, but was never-
theless "very confident'' the core was covered. 1 "
The evidence the TMI emergency command team
gave to support its belief the core was covered
was the spurious indicator of limited injection of
water from the core flood tanks. Although Stello
and Moseley questioned the basis for the utility's
lack of concern, neither of them asked for further
feedback from the utility, as Gilinsky had sug-
gested. 188 They had the opportunity to do so. Since
they had a direct line to the Unit 2 control room,
they could have raised the issue directly with Mil-
ler* (or for that matter any of the other utility
representatives), in keeping with Gilinsky's sug-
gestion that they speak with the plant superin-
tendent. Thus, the XRC left hanging the crucial
question of whether the core was uncovered.
" Stello. Moseley and other XRC officials said they believed the utility should open the block valve and leave it open.
causing pressure to plummet to the point where the decay heat removal system could be initiated. They did not know
that in the last attempt to depressurize. pressure had, on its own, stabilized at a point above that for low-pressure decay
heat removal. A more detailed discussion of this issue can be found in the staff report by the President's Commission.
"Report of the Office of Chief Counsel on the Nuclear Regulatory Commission." (6301
"~ See p. 141.
lra See p. 146.
147
The NRC Special Inquiry Group, after investi-
gating this particular matter, found :
There is no record of Stello having com-
municated this message [about super-
heated steam and an uncovered core] di-
rectly to the Unit 2 control room. . . .
(633)
As the tapes show, Stello did raise the issue with
the Unit 2 control room within minutes of his con-
versation with Gilinsky. The evidence suggests
that the Special Inquiry Group simply accepted
Higgins' recollection of the accident and was un-
aware of the contradictory evidence on the tapes. 169
Higgins' recollection was not supported by the
evidence uncovered by this Investigation.
NRC'S AFTERNOON STATUS REPORTS
About 5 :10 p.m., EMT members Case, Gossick
and Davis took part in a conference call with
members of the Subcommittee on Nuclear Regula-
tion and Senators H. John Heinz, III and Rich-
ard S. Schweiker of Pennsylvania. The Senators
had requested a briefing on the status of the re-
actor. Case summarized the various points of
view:
. . . The water level is above the core and
is showing in the pressurizer level which
is above the core. On this basis the com-
pany believes the core is covered and there
is no problem of further release of fission
products. It is not completely clear to us
that even though the core is covered there
may not be a steam bubble someplace in
the core which would result in inadequate
cooling to that portion of the core. We are
raising this question with the licensee,
suggesting that if this is still going on,
it might oe worthwhile to consider just
lifting the safety relief valve and blowing
the pressure down rapidly [depressuriz-
ing] in order to get this lower pressure
system on the line. The pressure has been
hung up around 500 pounds for the last
four or five hours. Slowly, slowly coming
down. But in the meantime, this portion
of the core may be overheating so that
is giving us some concern at this point in
time. (634)
Case described two possibilities : either the core
was completely covered or some small percentage
was uncovered. He did not point out that if the
core was uncovered, there were no direct means
at the plant for determining to what extent. (635)
The Senators asked if there was any need for
evacuation. Davis responded that offsite radiation
levels "do not at this time indicate evacuation."
(636) There was no mention of the uncertainty as
to the degree to which the core might be uncovered
and that that in itself was a reason for considering
protective action. This was because, as noted, the
NRC was focusing on actual radiation levels and
not on plant conditions in considering the need for
protective action.
The NRC's Role Discussed
The Senators then asked if the situation had
been stabilized. Case told them it had not and that
it might be a long time before stability would be
reached. The following exchange took place :
HART: Who determines the course of
action ?
CASE : The licensee.
HART: Under all circumstances?
CASE: Under all circumstances, unless
it gets to the point that if we think we
know enough here, which is very, very
difficult for us to conclude, that we ought
to tell them to do something. Now we
have direct communications through tele-
phone into the control building of Doth of
the units.
HART : How long will you wait ?
CASE : Pardon me ?
HART : How long might you wait before
you'd override them.
CASE : Well, it would be at least another
hour or so, I would think, Senator. (637)
Radiological Releases Reported
At about 5 p.m., the XRC issued its second
press release. It stated in part :
Low levels of radiation have been meas-
ured off the plant site. The maximum con-
firmed radiation reading was about three
milliroentgens per hour about one-third
mile from the site. At one mile, a reading
of one milliroentgen per hour was meas-
ured. It is believed that this is principally
direct radiation coming from radioactive
material within the reactor containment
building, rather than from release of
radioactive materials from the contain-
ment. (638)
The Information Is Wrong
The tape transcripts show that by 4 p.m.
IRACT had become aware of a 70 milliroentgen
per hour reading at the north gate to the plant.
They also show, when compared to Met Ed docu-
ments, that between 4 and 4:30 NRC inspectors
in Unit 1 had incorrectly transmitted as an onsite
measurement, a 50 milliroentgen per hour offsite
reading taken opposite the north gate on Route
441. (639)
' See Addendum 24, p. 160, for the text of Higgins' interview with the Special Inquiry Group.
148
With regard to the statement in the 5 p.m.
press release that the maximum confirmed offsite
readings were believed to be "principally direct
radiation," Sniezek, who was chiefly responsible
within IRACT for analyzing radiological infor-
mation, indicated that he did not recall having
discussed the question of direct radiation versus
releases in conjunction with the 5 p.m. press
release. 170 He said that if he had been asked, he
believed he would not have known whether the off-
site radioactivity was more the result of one than
the other : "I wouldn't know which one principally
it was." (640)
There are some direct measurements that can
serve as a check on whether a radiation dose is at-
tributable to direct radiation or to an actual re-
lease. If the radiation is the result principally of
direct radiation, then the dose measurements
should be somewhat constant in all directions from
the containment at a given distance, and they
should not change substantially unless the radia-
tion inside the containment changed substantially.
The information received by IRACT prior to
the issuance of the 5 p.m. press release was not
consistent with either condition. Rather, IRACT
was told that radiation levels inside the contain-
ment were constant ; the containment dome moni-
tor read 6,000 R/hr from 10 a.m. onward. Yet
between 2 p.m. and 4 p.m. dose measurements at
a given distance at the north gate went from 30
to 70 to 50 to 1 milliroentgens per hour. All this
was known to IRACT prior to 5 p.m. (641)
With respect to the general accuracy of the press
releases issued by the KRC on March 28, Joseph
Fouchard. Director of Public Affairs for the NRC,
stated. ". . . we were using the best information we
had in the Incident Center when we wrote these.
We were not trying to maximize or minimize the
situation. We were trying to tell it as we believed it
then existed." 171 (643)
More Incorrect Information
In the late afternoon IRACT also gave misin-
formation to Executive Branch agencies, including
the White House and the Department of Health,
Education and Welfare. Between 4 and 5 p.m.
Bernard Weiss, the IRACT Communications Offi-
cer at the Response Center and the person respon-
sible for briefing these organizations, reported to
Clark W. Heath of the Chronic Disease Division
at HEW's Center for Disease Control: "It was
really never a problem with regard to loss of water
and exposure to the core . . . there was never a prob-
lem of keeping the core covered." (644)
Between 5:30 and 6 p.m., Weiss discussed the
same issue with the White House Situation Room.
He reasserted that "there was never a problem with
regard to keeping the core covered." (645)
In fact, IRACT, from whom Weiss was receiv-
ing his information, had, by 4:30 pjn., spoken
with the Unit 2 control room. IRACT person-
nel were following up on the very concerns that
Weiss was telling the HEW and the White House
were not at issue that there were indications that
superheat in the primary system had been prevent-
ing adequate circulation through the core, leading
to possible uncovering.
Weiss, in explaining how he obtained the infor-
mation that he passed on to these agencies and in-
dividuals, said the data had gone through IRACT.
where it was evaluated, and then through EMT.
(646) "At some point it was said that we ought to
update the Commissioners on what is current at
this time." (647) He said he could not recall who
told him to tell the agencies that the core was not
uncovered.
Dudley Thompson, Weiss' superior, said that
Weiss "may not have been quite as current [on de-
velopments at the site] on a minute to minute
basis." (648)
Asked to explain the apparent discrepancies be-
tween what was known at the NRC and what was
being transmitted to other agencies through Weiss,
Case said he did not believe Weiss was "deliber-
ately misinforming" anyone. Case acknowledged
that Weiss' late afternoon report to the White
House Situation Room "simply wasn't accurate." 1
(649) His explanation was that the report resulted
from the normal confusion that arises in an emer-
gency. (650)
THE NRC FAILS TO FOLLOW UP
On March 28, the NRC never did explore the
need for evacuation or take steps to override the
licensee with respect either to its diagnosis of the
severity of the accident or to actions that should
be taken to regain control of the reactor. 172 Accord-
ing to Commissioner Bradford, one reason evacu-
ation was not addressed was that the Commission
"simply did not have the information on what was
going on inside the reactor." (651) While that cer-
tainly was the case in the morning, it was not so
in the afternoon.
1TQ See p. 158.
71 The NRC's Special Inquiry Group concluded : "To anyone acquainted with reactor physics, the idea of a contain-
ment building so full [of] radioactivity that it is penetrating those 4-foot concrete-and-steel walls with enough intensity
to be picked up by monitors more than a mile away well, it is not only grossly in error, but ridiculous in retrospect."
175 It is now believed that if the utility had pursued the strategy of rapid depressurization favored by the XRC, an
even more serious condition could have developed.
149
Chairman Gilinsky explained to the Subcom-
mittee why he took no steps to initiate discussions
on protective action with the other Commissioners
or with the State after he learned of Stello's con-
cern that the core was uncovered :
Let me tell you what was on my mind.
The comparison that I made continually
was with the temperatures I'm familiar
with from the rules on emergency cooling
systems, which require that tempera-
tures in the reactor core stay below ap-
proximately 2,000 degrees, 2,200 degrees,
during the course of [a] loss of coolant
accident. This is based on the fact that
oxidation of the cladding becomes rapid
at about 1,600 degrees. So mentally, I
was making this sort of a comparison,
throughout the day in fact. And none of
the temperatures that I had heard ap-
proached anything like those numbers.
Now, I must say that I asked how we
could be getting fuel failure if we were,
in fact, nowhere near such temperatures,
and I remember the response, I don't re-
member who gave it, that one could get a
certain amount of fuel failure, pin holes
in [the] fuel, if the fuel [saw] something
like a thousand degrees for sometime.
So clearly in my mind I had some sort
of picture of either pockets of steam or
some form of inadequate cooling. But it
did not, to my mind, at least at that point,
call for further steps with regard to evac-
uation. . . . (652)
Gilinsky said he recalled that the discussions
centered on the extent of fuel failure, which was
then estimated to have been one percent, rather
than on core uncovering. (653)
This figure of one percent fuel failure, which
proved to be wrong and which suggested far less
serious conditions than existed, was used repeat-
edly not only in statements to the press that eve-
ning, but in testimony by the Commissioners the
following day before the House Subcommittee on
Energy and the Environment. (654) It is unknown
where the figure came from.
One possible source was a previous NRC esti-
mate that certain "design basis" accidents would
result in a failure of one percent of the fuel,
which would in turn produce an iodine spike. 173
(656)
Yet, even at this point the NRC had evidence
showing that the figure of one percent was incor-
rect. As early as 4 p.m. on Wednesday some
NRC staff had ruled that figure out as inaccurate.
An NRC official in IRACT, and another in the
main NRR offices, had discussed some NRC calcu-
lations based on the primary coolant sample the
utility had drawn and analyzed earlier in the
morning. It showed more radioactive iodine than
would be found in the coolant as a result of an
iodine spike. (657) It was evidence of fuel failure
greater than one percent and suggested greater
damage to the core.
The NRC has since estimated that during the
first day, over 90 percent of fuel had failed, that
the entire inventory of radioactive iodine in the
core was released to the coolant, and that the ge-
ometry of the core was disarranged. (658)
COMMISSIONERS' ROLE IN RETROSPECT
The Commissioners played a relatively minor
part on March 28, notwithstanding their pre-
scribed policymaking function. As Eisenhut com-
mented, "To the best of my knowledge [the Com-
mission] really played no firm policy direction
role on Wednesday." (659)
The limited participation of the Commissioners
was not surprising, in retrospect. The presumption
had always been that accidents would be of such
short duration, there would be no time for the
Commissioners to become actively involved. 174
They were clearly unprepared to do so.
Second, neither the Commissioners themselves
nor the senior emergency response staff saw the
Commissioners as having an operational role.
IRACT's Stello noted,
. . . my view was the decision [to direct
the licensee] would be made by EMT if
needed, and I didn't think very much
about the Commission. If we needed to
decide something, my view was that [a]
decision would be made and inform the
Commission rather than asking. (660)
Case, an EMT member, told Special Investiga-
tion staff that he felt the Commission members
should ". . . keep out of it ... I don't think the
five-man body, whomever they may be, is the type
of organization you want in an emergency." (661)
He said the Commissioners' role was "delibera-
tive" and that they should not be involved in
handling emergencies. (662)
The Commissioners themselves stated that it
was appropriate for them to rely on the emergency
3 "Design basis" accidents are hypothetical events analyzed by the NRC in terms of plant response and of safety
features required to handle the accident. Some design basis accidents could result in damage to the Zircaloy cladding
on the fuel rods. Because of the damage, some radioactive iodine, a fission product normally contained by the cladding,
would be released to the coolant. The release would show up as an increase in radioactive iodine, referred to as au
"iodine spike." The size of the spike would be indicative of the amount of failed fuel. (655)
m See "Prior to the Accident," pp. 82-83.
150
response organization for technical decisions.
Commissioner Ahearne commented :
As far as the issue of what is the role
of a Commissioner during emergency re-
sponse, my understanding of it prior to
and certainly during [the accident] was
that the way the NEC system was de-
signed was for the senior technical people
in the agency to be responsible for moni-
toring and taking whatever action might
be necessary as far as the technical issues.
(663)
Commissioner Gilinsky noted :
generally speaking, the technicaJ,
minute-by-minute decisions and recom-
mendations have to be handled by our
staff. And the Commissioners have got to
deal with things that are more general in
nature . . . but the technical questions have
got to be examined by the staff, and it is
they who have to be in direct touch with
the licensee as well as counterparts in the
State. (664)
Nevertheless, there were actions the Commis-
sioners 'believed, in hindsight, they should have
taken. As noted, Chairman Hendrie spent the first
day of the accident at a hospital with his daughter
and was not heavily involved in events on the first
day. He subsequently said that "it wasn't very ef-
fective for me not to be there ; I should have gone
to the response center." (665) Commissioners Ken-
nedy, Bradford and Gilinsky thought the issue of
evacuation would have been formally addressed
by the Commission on Wednesday morning had
they had information available to the utility.
(666)
STABLE CONDITIONS ACHIEVED
At about 4:30 p.m., Jack Herbein had arrived
at Unit 2 after briefing the Lt. Governor in Harris-
burg. He decided that because of the unsuccessful
attempts to depressurize and bring on the decay
heat removal system, the control room operators
should repressurize again, with the aim of restart-
ing the reactor coolant pumps. Herbein discussed
the matter with GPU Vice President Robert
Arnold, who was at GPU headquarters in Read-
ing, Pennsylvania. Arnold concurred. (667)
At some point between 5 and 6 p.m., Her-
bein, concerned that the core might not be covered,
ordered the operators to stop depressurization,
raise reactor system pressure and try to start the
coolant pumps. (668) At 6 p.m. Higgins informed
XRC headquarters of this strategy. (669)
This time repressurization was successful. The
bubbles in one loop of the primary system col-
lapsed, and one of the pumps was started at about
7 :50 p.m. It forced coolant through the core and
allowed heat to be removed through the steam gen-
erator. Thus, approximately 16 hours after the
accident started, circulation through the core and
heat removal through a steam generator were
achieved. Relatively stable plant conditions were
finally established. The immediate crisis had
passed.
POTENTIAL FOR GREATER SEVERITY
A considerable amount of the analysis since the
accident has focused on whether different se-
quences of events could have posed greater danger
to residents of the surrounding community. The
answer depends, to a great extent, on the prob-
ability that the accident could have resulted in the
melting of the reactor core or in offsite releases
of hazardous levels of radioactivity.
The actions of plant operators and managers
did lead to substantial uncovering of the reactor
core. Calculations done for the NRC Special
Inquiry suggest that the core would have begun
to melt within an hour after the block valve
was closed if plant personnel had failed to close
it and continued to limit the flow of high pres-
sure injection. (670)
THE OUTCOME OF CORE MELTING
The President's Commission also looked at al-
ternative sequences of events. (671) It concluded:
No single additional operator action or
equipment failure that is tied to the
actual sequence of events at TMI would
have led unequivocally to large scale
fuel melting throughout the core or sig-
nificantly larger releases of fission prod-
ucts to the environment. (672)
Contrary to what a recent report by the
House Subcommittee on Energy Researcn and
Production concluded, this finding leaves open
the possibility that multiple incorrect operator
actions minimal or no high pressure injec-
tion, accompanied by no heat sink and con-
tinued let-down could have produced those con-
ditions. (673) Indeed, the staff of the Presi-
ent's Commission identified four possible
"serious cases" in which large-scale fuel melting
could have occurred. (674) However, when they
studied the radiological consequences of these
four cases, they concluded that even in those
cases, containment integrity probably would not
have been violated. (675) The President's Com-
mission also found that there would have been
151
no substantial radioactive releases from the plant
even if the core had melted through the contain-
ment floor because the bedrock foundation prob-
ably would have contained the radioactive debris.
(676) It also concluded that the release of fission
products would have been greater than actually
occurred, although not by a "large factor." (677)
PROBLEMS WITH THE ANALYSES
There are a number of problems with the "what
if" analyses of the NRC Special Inquiry Group
and the President's Commission, including the
degree of uncertainty that attends all such studies
of hypothetical events. The first implicitly as-
sumed that a core melt following initial un-
covering of the core in the morning would have
been accompanied by isolation of the containment
to prevent the escape of radiation. (678) In fact,
the utility consistently bypassed containment iso-
lation throughout the day in order to use the
make-up and let-down systems. (679) The let-
down system was leaking; this leakage was pri-
marily responsible for the actual releases of radio-
activity outside the containment and into the aux-
iliary building and then to the environment. Had
a core melt occurred while containment isolation
was bypassed, the releases to the environment
through the auxiliary building pathways would
have been greater during the period between the
beginning of the melting and the eventual rup-
ture of the reactor vessel.
An important issue that is unaddressed by these
studies is whether the operators, based on the
available i nstrumentation, would have realized
that the core was melting down, and whether melt-
ing would have required protective action at a time
when no one was prepared for it.
Of these analyses, only the President's Com-
mission addressed the leaking let-down system as
a pathway for releases of radioactivity from the
containment building. It did so, however, in
terms of what actually occurred, rather than in
relation to a hypothetical sequence of events lead-
ing to a core melt. The President's Commission
concluded on this basis that greater, but not sig-
nificantly greater, releases would have resulted
from a core melt at TMI. (680)
The President's Commission relied on a two-
part qualitative argument in reaching this finding.
First, it had studied the actual behavior of the
radioactive iodine during the accident and its re-
lease through the auxiliary building. Its prelimi-
nary finding was that little radioactive iodine was
released, much less than would have been pre-
dicted based on the scientific literature. (681)
The staff of the President's Commission issued
the caveat that "anyone who thinks he thoroughly
understands why iodine did what it did during the
accident is following a simplistic approach. . . ."
(682) They had been unable to make quantitative
calculations of the likely magnitude of releases
accompanying a hypothetical core melt. (683) In
part, this was because they were uncertain about
the chemical conditions in the reactor coolant sys-
tem at TMI and how those had affected the actual
releases of iodine, although their studies showed
that the limited releases related closely to chemical
conditions (such as the ph of the coolant) . To con-
clude on the basis of what happened at TMI that
limited releases would have resulted from a hypo-
thetical core melt assumes that the same (un-
known) chemical conditions would be present.
This in turn assumes that during a core melt oper-
ators would have access to a reactor coolant sam-
ple and would again fortuitously misread the
boron concentration, as the TMI operators did a
misreading that had led them to alter the chemis-
try of the coolant, coincidentally resulting, it now
appears, in lower releases. (684)
Second, at TMI the inventory of volatile fission
products (685) was mostly released to the coolant
during the initial uncovering of the core. Given
that fact, the President's Commission argued that
the release of the remaining fission products that
would occur with a melt would not have produced
significantly larger releases than actually oc-
curred. (686)
The Special Investigation staff found that this
argument neglected the uncertainties of condi-
tions in the reactor coolant system. The hypo-
thetical cases analyzed by the President's Com-
mission did not adequately consider many factors
that need to be addressed in predicting whether
melting would have caused larger release rates
through the auxiliary building and, if so, how
large those might have been.
Staff of the President's Commission used the
Commission findings to assert that the "health ef-
fects" accompanying a core melt would have been
unobservable. (687) This conclusion does not take
into account either the accompanying uncertainties
or the psychological impact the accident had on
the local community.
The report of the House Subcommittee used the
various "what if" analyses to conclude that "there
was always a reasonable margin of safety during
the accident at TMI." (688) For the reasons cited
above, it is difficult to reach such a conclusion with-
out postulating operator actions of doubtful
probability and without setting aside the issue of
psychological impact.
152
ADDENDA TO CHAPTER 7
Addendum 1
Ros? was referring to a precaution contained
in TMI-2 operating procedure Xo. 2101-1.1. It
states :
1.2-01 Absolute maximum pressurizer at
any time the reactor is critical is 385
inches, [emphasis added]
NOTE : This water level is the maximum
RCS [reactor coolant system] inventory
used in the safety analyses for reactor
building overpressure following a
LOCA. It is also the maximum level at
which the system can accommodate a tur-
bine trip without causing the pressurizer
safety valves to open.
1.2-04 The pressurizer must not be filled
with water to indicate solid water condi-
tions (400 inches) at any time, except as
required for the system hydrostatic tests.
(689)
Addendum 2
Zewe also noted that at one point he thought that
some of the circuit breakers for the heater in the
pressurizer had blown and that, as a result, the
pressurizer had ". . . lost some heater capacity
and we just couldn't recover pressures as fast as
we should." 1T5 (690) He asked one of the control
room operators to have an auxiliary operator
check the heaters. (691) However, according to
Zewe :
I'm not sure whether he did check it and
report to the operator. The operator
didn't tell me. I don't remember any de-
tails of it. and I really didn't pursue it
any further. (692)
Addendum 3
The plant 's emergency procedures gave them no
clear guidelines for making that decision, since
the procedures assumed that RCS pressure and
pressurizer level would trend in the same direction
during a LOCA. In Ed Frederick's words:
A combination of high T h I17Sl and low
pressure and a full pressurizer was
enough. . . . We might not as well have an
emergency procedure book once you see
something like that. There's nothing that
you can figure out from that point. (693)
In Frederick's words, the high pressurizer level
and low RCS pressure were ". . . confusing indi-
cations that don't dictate anv particular course of
action." (694)
Addendum 4
The operators needed to open a valve that would
allow feedwater to bypass the malfunctioning
condensate polishers, (695) but the remote con-
trol switch in the control room was not working.
(696) In addition, the water level in the condens-
er hot well was excessively high, 1 " which, if not
corrected, would preclude use of the main feed-
water system. (697)
Addendum 5
Bryan said that when the operators first con-
cluded that the rupture disc had blown. ". . . we
figured the safety lifted on the pressurizer and
blew the rupture disc you know, just overpres-
surized it.*' (698) However, in another interview
Bryan contradicted himself. ". . . [I] never
thought that the code safety valves opened/' (699)
Bryan added that since he knew the reactor cool-
ant system pressure had not gone over 2,355 psi,
he did not believe the code safety valves had
lifted. 178 (701)
Addendum 6
Zewe attributed the symptoms to excessive water
in the drain tank :
I knew at this point that we either had the
RC [reactor coolant] drain tank's relief
valve open or the rupture disk had
blown (702)
According to Zewe, an alarm indicating that the
sump pumps were running had been activated at
about 4:08, 8 minutes into the accident, but the
operators did not notice it because of the backlog
on the computer (703) which was printing out
171 Pressurizer heaters are normally used to enlarge the steam bubble in the pressnrirer by heating the water in
the pressurizer. turning some of it to steam. This in turn increases the pressure in the system.
! *Th is hotleg temperature.
' The condenser hot well is where steam condenses after passing through the turbine. If the level is high, it will
inhibit further condensation of steam.
" The code safety valves are designed to lift at 2,435 psi. (700)
153
- 80 - 11
the alarms. In Zewe's opinion, "I believe that the
indication [for the reactor coolant drain tank] we
have available in the control room is insufficient.
. . ," 179 (705)
Addendum 7
According to Zewe,
I felt certain that the water that was
going into the reactor building sump was
from the RCDT [reactor coolant drain
tank] and also that's also where the [in-
crease in containment] pressure was
from. (706)
Implicit in Bryan's comments in one interview
is that he thought the same things. He said that
at one point the operators thought a steam line
was leaking into the containment, but that ". . . we
went right back to the assumption that we had a
rupture disc blown in the drain tank." 18 (707)
Addendum 8
Operators, in using emergency procedures, may
read from them or refer to them from memory
(operators are required to memorize the pro-
cedures). However, according to Scheimann, op-
erators do not necessarily refer directly to the
procedures, especially in the early stages of an
accident. Then they tend to focus on the symptoms
and what response might correct it :
. . . [The operators] don't just sit there
and say, "Oh, mercy sakes, I got a loss of
pressurizer level there", or "Mercy sakes,
look, pressurizer pressure is going down.
I have got to refer to emergency proce-
dure blah-blah-blah." Your train of
thought, just doing work like that in a
situation of that nature, you just see a
symptom and you try to correct for what
that symptom's problem is. ... (708)
Frederick described how operators determine
which procedure is relevant:
FREDERICK : The thought process is ac-
tually not one of trying to eliminate each
emergency procedure that exists. What
you are trying to do is assemble a certain
amount of symptoms that you can apply
to an individual emergency procedure.
Question : But how do you know which
ones to look for ?
FREDERICK : We don't look for particu-
lar symptoms; you wait for them to be
evident, and you make a list of them in
your mind, and you try and decide which
of those is important and how they relate
to the emergency procedures. In other
words, not only the symptoms but the
order in which they appear will steer you
to a different emergency procedure. (709)
Frederick added,
Usually at the beginning of a transient
like that the emergency procedures that
you use later on are not related to the
original problem. This is exactly what
happened to us. We had a loss of feed-
water and many of the emergency pro-
cedures we might have used were not at
all related to feedwater. And you had to
pick up the symptoms along the way.
(710)
Addendum 9
The control room personnel considered the first
two scenarios a steam line rupture or a primary
tube to secondary system leak to be plausible be-
cause of problems in the "B" steam generator.
They had observed both low pressure and high
level on the secondary side of the generator.
At 5:27 a.m. they isolated the "B" steam gen-
erator. 181 (711) The control room personnel dif-
fered as to why. Most recalled that they explicitly
considered the possibility that the accident in-
volved a steam leak. Faust, Zewe, Bryan, Schei-
mann and Kunder all thought a steam leak could
have been contributing to the rising pressure in
the containment. (712)
Zewe, Kunder and Faust said that when the
pressure in the "B" steam generator dropped to
300 psi less than the pressure in the "A" steam
generator ("A" steam generator pressure was
about 1,000 psi) at about 5:30 a.m., they decided
there was a rupture of some kind in a steam line.
(713) Zewe, Scheimann and Kunder said the "B''
steam generator was isolated because the control
room personnel thought there was a steam line
break. (714)
Faust, Scheimann and Bryan said, in addition,
that the operators believed the problem was a pri-
mary to secondary tube leak. (715) In Faust's
words :
I kept pushing myself that we had, first
of all, a steam generator tube leak simply
because I had an increasing level in the
"B" generator, and I could not terminate
it. It had to be coming from somewhere.
(716)
179 However, in another interview Zewe said the sump pumps usually come on about once each 8-hour shift to remove
condensation from the sump. (704) Thus, the operators might not have interpreted the sump pump alarm as an unam-
biguous sign of an unusual amount of water.
180 A key symptom of a leaking steam line inside the containment is a rise in containment pressure. Bryan's state-
ment implies that the operators first thought the rupture of the drain tank was responsible for the rise in pressure, then
thought that the steam leak was responsible, and finally attributed it again to the drain tank.
181 Isolated means they stopped flow of feedwater to It.
154
Brvan said they isolated the generator because
they thought it had a tube leak. 182 (717) Faust said
he believed there was both a tube leak and a steam
line break, although "I mainly wanted to isolate
the -B' generator because I thought it has a tube
leak." (718)
Faust said that he also thought there may have
been a break in the emergency feedwater line as
well. (719) a result of his rapid initiation of emer-
gency feedwater How eight minutes into the acci-
dent! 183
The evidence indicates that Zewe was the only
other person aware of this hypothesis. Faust noted
that "[Zewe] didn't fully agree with me on it"
( 7-2-2 ) Faust also said he believed the break could
have been the source of the water in the contain-
ment sump. (723)
Addendum 10
The control room personnel found the proce-
dures to l>e vague or unclear and incomplete.
TMI-2 Emergency Procedure 2202-1.5, "Pres-
suri7A'r System Failure." listed the symptoms for
a failed PORV. Zewe said that the control room
personnel referred to this procedure during the
accident, although he noted that "we did not spe-
cificallv pull out that procedure until later localise
we did' not suspect that we had the relief valve
problem." (7:24)
Symptom 2 of the procedure "RC System. Pres-
sure in Mow 2205 psig and RC-R2 [PORV valve]
fall* to clone" implies a tautology: that PORV
failure ("RC-R2 fails to close") is a symptom of
PORV failure. Further. Frederick interpreted the
procedure to be referring to the PORV position
indicator light in the control room, rather than the
PORV itself. (725) Since the absence of the light
indicated a closed valve, he did not consider the
symptom to be applicable. (726) Zewe. too, be-
lieved it to have closed because of the absence of
the light :
... we have a red light for the valve
whenever it lifts of course that was still
out and I didn't realize it ... was still
hung open. . . . (727)
Symptom 4 : "The RC drain tank pressure and
temperature are above normal on the control room
radwaste disposal control panel 8A" was reviewed
bv the control room personnel several times during
the first two and one-third hours of the accident.
The procedure does not mention that pressure in
the reactor coolant drain tank will rise steadily
until the rupture disc bursts, at which point pres-
sure will return to normal. The symptoms referred
only to conditions which would exist immediately
after the PORV became stuck open.
In addition, the procedure did not discuss how
the water level in the tank would behave if the
PORV were to stick open. While actual water
levels could not be specified by the procedure be-
cause the drain tank collects leakage from many-
places in the primary system, how the level would
change could have been addressed.
The operators had focused on the water level
in the drain tank in their efforts to diagnose the
accident. Zewe said he concluded, given the ab-
sence of a procedure on water level and without
data on trends, that the low level in the drain
tank indicated the PORV was not venting water
into the tank and was therefore closed. (728)
Overall, the procedure described the various
symptoms too generally. As a case in point, it
stated that temperature and pressure would be
"above normal." symptoms so broad that they
also applied to those produced by a normally
functioning PORV during a reactor trip.
With respect to the role emergency procedures
played in the operators' decision to throttle HPI,
Zewe said that when the operators manually acti-
vated the first HPI pump, he referred to TMI-2
Emergency Procedure No. 2202-1.3, "Loss of Re-
actor Coolant/Reactor Coolant System Pressure.''
(729) Sections 3.0 to 3.2.2, Part A, entitled "Leak
or Rupture Within Capability of System
Operations." (730) These sections describe the
steps operators should follow when they manually
initiate HPI. The last one (3.2.2) to which Zewe
said he progressed directed operators to throttle
HPI if the level of water in the pressurizer went
over 220 inches :
Bypass the SAFETY INJECTION by
DEPRESSING the Group Reset Push-
buttons & "THROTTLE" MU-V16A/B/
C/D as necessary to maintain 220" pres-
surizer level and not exceed 250 GPM/
HPI flow leg. (731)
Section 3.2.5 on the next page of the procedure
contains the following warning:
CAUTION: Continued operation de-
pends upon the capability to maintain
pressurizer level and RCS [reactor cool-
ant system] pressure above the 1640 PSIG
[pounds per square inch gauge] Safety
Injection Actuation Setpoint. (732)
1K A tube leak would involve a break in one of the many small pipes within the generator through which heat from
the coolant is transferred.
110 Faust said that when he opened the block valves on the emergency feedwater line, a microphone near the main
steam piping picked up the sound of ". . . cold water going down a hot pipe and hitting the hot-steaui generator." (720)
Faust noted that : "I thought then there was a break in the emergency feed line possible, not that the line sheared off,
but a break somewhere due to thermal shock." (721)
155
Zewe missed the caution. He said,
I never went that far [in reading the sec-
tion]. I was still at the point of the pro-
cedure under the previous page of trying
to throttle high-pressure injection flows
to maintain levels. (733)
Zewe did not say why he stopped, but implied
that, since he was unable to reduce pressurizer
level to 220 inches, he never went beyond that step
of the procedure. (734)
Frederick commented on the difficulty of writ-
ing emergency procedures:
The tough part about any emergency
procedure is writing comprehensive
symptom type statements that'll get you
started on a procedure. And it's hard to
anticipate any kind of or all of the situa-
tions that would start you on a procedure.
Symptoms have to be general, they have
to be general and specific at the same time.
You have to try and accomplish a wide
number of circumstances, but they have
to use specific indications to get you
start fed]. So it's a tough assignment. . . .
(735)
Addendum 11
Zewe stated that at one point he was referring
to Part A of the LOCA emergency procedure
2202-1.3. When the Engineered Safety (ES) sys-
tem was activated at two minutes into the acci-
dent, that part of the procedure was no longer
applicable. Instead, the relevant part was "B,"
"Leak or Rupture of Significant Size Such that
Engineered Safety Features Svstems Are Auto-
matically Initiated." Zewe said he did not refer
to that part of the procedure during the accident
because he did not believe the activation was the
result of a LOCA. Had he followed Part B, as
directed by the procedure, he would have known
to leave the HPI pumps running until he could
turn on the low pressure injection pumps to cool
down the reactor. (736)
Addendum 12
Zewe said that cooldown by natural circulation
was discussed in the TMI-2 emergency procedure
covering loss of off site power. 184 (737) However,
in a group interview witli Scheimann, Faust and
Frederick, he commented that the procedure
contained
... no real detail on what to look at or
how long to look at it ... or how long
you'll have to wait before you start to
see invalid indications one way or the
other. (738)
Addendum 13
The first, and possibly the second, reading was
obtained by Bryan at Zewe's request. (739)
There is reason to question whether it was
Bryan who called up the readings at 5 :20 a.m.,
80 minutes into the accident. He recalled,
I checked the temperatures at least
twice, maybe three times within the
first couple of minutes, well, within tin-
first half hour that I was there. And
each time all three of them [the PORV
and the code safety valves] indicated
I forgot the numbers, but they were
within 15 degrees or something like
that. . . . (740)
However, many operators had poor recollec-
tions of when events occurred during the acci-
dent. Bryan's comment that the temperatures of
the three valves were within about 15 of each
other is consistent with the readings taken at
4:24 a.m., 24 minutes into the accident, and not
with those taken at 80 minutes. At that point
the PORV was about 65 hotter than the code
safety valves. This discrepancy is significant,
since Bryan claimed to have been focusing on
the temperature difference between the PORV
and the other valves. Either (1) someone other
than Bryan took the reading at 80 minutes, or (2)
Bryan misread the reading.
Mehler asked for the last set of readings. From
them he concluded the PORV was open. (741)
Addendum 14
In interviews after the accident. Bryan recalled
only that the temperatures of the PORV and
code safety valve discharge lines were within about
15F of each other. (742) Since all three valves
had elevated discharge line temperatures, he con-
cluded the PORV was not stuck open "Fbe]cause
the other two are indicating the same." (743)
Bryan noted. "I know I looked at the indications
for the valves and it indicated closed.'' (744)
When he was asked why he did not suspect the
PORV was open, he replied. "It indicated shut.
All three relief valve temperatures were approxi-
mately the same." (745)
Comparing the temperatures of the PORV and
code safety valves for diagnosing PORV failure,
was not discussed in the emergency procedure.
Rather, operators were to consider only the tem-
perature of the PORV discharge line.
Bryan did not state why he used this incorrect
diagnostic method. Other operators have said,
however, that during normal operations, differ-
ences in temperature between the PORV and
'TMI-2 Emergency Procedure No. 2202-2.1.
156
the code safety valves could be used to determine
whether a valve was leaking. (745) Bryan may
have assumed the same principle applied in an ac-
cident in which a valve stuck open.
The Special Investigation staff believe, on the
contrary, that in sucli a situation heat would be
transferred to all the discharge lines, so that, at
least initially, the valves would have similar
temperatures.
Zewe has conflicting recollections about what
Bryan told him about the discharge line tempera-
tures and how he interpreted them. In one inter-
view. Zewe said Bryan checked the temperatures
at 4:24; 24 minutes into the accident, and that
. . . they didn't look abnormally high since
the electromatic [PORV] had lifted. It
was about 228 or 230 degrees and they had
been running about 170 to 180 so I figured
it was still warm from when it lifted.
(747)
This statement suggests that Zewe was looking
at how much the temperature was elevated above
normal. It also suggests that he discounted the pro-
cedure's warning that a temperature of over 200
in the discharge line was a symptom of PORV
failure, both because he was aware that tempera-
ture in the line prior to the accident was elevated
and that the valve had lifted.
In another interview Zewe implied that he,
like Bryan, relied on a comparison of the tempera-
tures of the PORV and code safety valve discharge
lines:
I ... had him check the discharge tem-
peratures of the relief valves, and he
said you know the RCRV 2 [PORV] is a
little* high, about 30 higher. (748)
It is possible that Bryan never told Zewe what
the temperatures of the three valves actually were,
instead noting only that they were within 30
of each other, and that Zewe subsequently con-
fused the 228F reading which Mehler obtained
at 5:17. 2 hours and 17 minutes into the accident,
with what Bryan told him at 4 :24. 24 minutes into
it. This would explain Zewe's previous statements
that the PORV temperature at 24 minutes into
the accident was about 230F. although subse-
quent analysis has shown it was 285F.
Addendum 15
Although Mehler did reach the right conclu-
sion about the steam and was on the right track
concerning the PORV. he had not used the steam
tables. Like the others, he did not deduce the steam
was superheated and that the core was uncovered
and being damaged. He later explained :
. . . I T p until [the radiation alarms came
in at 6:40 a.m.]. I thought [that they
just] had steam voids in the hotlegs . . .
That was the only place we had them
which led me to believe we had not mv
covered the core at that time. I did realize
we had problems and fuel failure when all
the alarms came on. Until that point
there was no indication we did have fuel
damage. (749)
Addendum 16
Logan was perplexed by the source and inter-
mediate range monitors :
... I might add, at the same time that we
lit the pump off we had an indication of
a count rate increasing. We had at the
same time received a chemical analysis
indicating that our boron . . . was lower
than we had anticipated . . . There
were several abnormal indications going
through there. (750)
Addendum 17
Benson's description to Special Investigation
staff was :
. . . I basically looked at the reactor cool-
ant temperature; the [hotleg] was
pegged [high] the [coldleg] was 130 to
140 degrees : it was just the opposite . . .
I look at the pressure; it was down. I
noticed there was no flow ; all the reactor
coolant pumps were off ... I looked at
the start-up [source] range and . . . the
intermediate range . . . and they had
both [come] down to what appeared to
be normal after they made that one pump
start a little earlier.
So I figure I would see what the incores
[the incore neutron detectors] read.
When you're below 15% power the com-
puter won't do certain calculations, one
of them being the incores . . . That
wasn't the case. There were some of them
printing out full scale ... I noticed that
the ones that were printing offscale were
basically the hot channels or ... [fuel]
assemblies that you would expect to be at
the highest [neutron] flux ... It seemed
like the information was pretty good be-
cause it was actually showing the correct
assemblies I would expect to have the
highest decay heat, but I couldn't believe
they were offscale ... I assumed one time
we had a void go through the core. . . .
(751)
Addendum 18
Two other factors contributed to the contin-
uing difficulty the XRC had with internal com-
munications on March 28. both outgrowths of
the inadequacy of the emergency response plan-
157
ning and implementation of NEC's emergency re-
sponse program. One was the actual flow of in-
formation between TRACT and the EMT on
March 28, as compared with the intended flow, as
depicted in the NRC Headquarters Incident Re-
sponse Plan. Second was the failure of the Reac-
tor Operations Inspection 185 implementing pro-
cedures, in effect during this incident, to include
staff of the Office of Nuclear Reactor Regulation
(NRR) as part of the emergency response orga-
nization.
Although the Response Plan specified that in-
formation between IRACT and the EMT would
go through a predesignated liaison, on March 28
no set pattern was followed in the transmittal of
data. The EMT received briefings from a number
of IRACT support staff and team members. (752)
Further, different EMT members would confer on
their own with various members of IRACT, and
"nearly all of the communication that took place
back and forth between IRACT and EMT was
verbal." (753)
Edson Case, the representative NRR had as-
signed to the EMT on March 28, told Subcom-
mittee staff that the EMT never had any sort of
formal meeting and that he received his informa-
tion primarily from his NRR counterpart on
IRACT, Victor Stello. (754)
Case's comment reveals an even broader prob-
lem the separation between NRR and I&E staff
on the EMT and IRACT. Although, as noted, the
ROI implementing procedures did not call for
participation by NRR, the NRC Headquarters'
Response Plan did, and several NRR personnel
were assigned to the two teams. The incident re-
sponse organization in turn functioned to some
extent as though there were two separate organi-
zations one of NRR staff and one of I&E. For
example, individual EMT "team members were
speaking to members of their respective organiza-
tions to obtain updating information on particu-
lar items of concern to them." (755)
There were also, in effect, two IRACT's one
under the Director of IRACT. who was from
I&E, and one under the IRACT member from
NRR. One illustration of this division involved
two IRACT Support Groups Plant Systems Ef-
fects and Radiological and Environmental Ef-
fects. James H. Sniezek, Director of I&E's Divi-
sion of Fuel Facilities and Material Safety In-
spection, headed the radiological group, as speci-
fied in the implementing procedures. However,
according to Darrel Eisenhut, Deputy Director of
NRR, Brian Grimes, an TRACT support staff
person assigned by NRR, transmitted the radio-
logical information received at the Response Cen-
ter to a reactor systems team located in the build-
ing where NRR had its main offices. (756)
There is evidence that this second team was dis-
tinct from the IRACT radiological group staffed
by I&E personnel and headed by Sniezek. For ex-
ample, the physical layout of the Center provided
a station for the radiological group's activity, but,
according to Sniezek, he and Grimes did not share
that location :
Question: Mr. Grimes was working
with you wasn't he, on the radiological
effects?
SNIEZEK : We were not working di-
rectly in the same physical location. . . .
(757)
Another statement by Sniezek suggests that he
and Grimes were not actually working together:
Question: Was Mr. Grimes working
with you on March 28 ?
SNIEZEK: He was in the incident re-
sponse center.
Question : Was he following radiologi-
cal information ?
SNIEZEK : He was involved somewhat
in following radiological information
also. (758)
Who was heading the TRACT Plant Systems
Effects Support Group was an open question. Ac-
cording to Grimes, it was Stello :
I think in effect Mr. Stello was . . . leading
the [plant] systems evaluation as the most
knowledgeable person in the field, and
looked to me for radiological evaluations,
as the primary source of [those kind of]
evaluations, and the I&E function was
communication and collection of infor-
mation, as had been planned. (759)
However, according to TRACT team members
Harold Thornburg and E. Morris Howard, both
I&E Division Directors, Plant Systems Effects
was under the direction of Norman Moseley, Di-
rector of TRACT. (760) Moseley, on the other
hand, said the group was headed by IRACT's
Technical Coordinator, Samuel Bryan. (761)
Bryan in turn stated that while he was following
operational and plant systems issues, he was not
the Technical Coordinator. He thought Edward
Jordan, Assistant Director for Technical Pro-
grams, had that job. (762)
Bryan at times served as back-up for the field
communicator holding the phone to the site open,
receiving information and asking questions over
the open line. (763) It would seem impossible for
him to have been coordinating the activities of
the two support groups. Nor is it apparent that
" The Division of Reactor Operations Inspection is within the Office of Inspection and Enforcement. Its implementing
procedures were to be applied in the event of an incident involving plant operations. Thus its personnel were designated
as support staff for the Incident Response Center. See "Prior to the Accident," p. 80.
158
Sniezek was reporting to Bryan, and, as noted, the
XRR representatives in both groups tended to
deal primarily with one another and with their
colleagues at XRR headquarters. (764)
There is no evidence showing who, if anyone,
was acting as Technical Coordinator.
The Reactor Operations Inspection implement-
ing procedures were unclear about the role of the
Technical Coordinator. They required that the two
support groups coordinate the agency's entire re-
sponse in their areas of responsibility. (765) How-
ever, the procedures did not stipulate how that was
to be done. For example, they did not assign any-
one responsibility for the task ; the procedures only
mentioned that members of the two groups should
report to a Technical Coordinator. (766)
Addendum 19
Miller noted that except for those items on the
checklist for initial notification, the utility did not
discuss matters like coolant level with State
people :
Question : I am sure they were commu-
nicating radiological information to the
State. I am not sure that the information
regarding the plant status was being
transmitted . . .
MILLER : . . . There is ... a checklist for
plant conditions in the Emergency Plan
... it is geared to talking about things
that need to be opened all during an acci-
dent makeup pumps, diesel generators.
I think if you look at that we would have
probably conveyed that. I am not sure we
would have conveyed the discussion you
and I are having about the core level, the
core flood tanks. (767)
Addendum 20
In answer to the question, "to your knowledge,
none of the operations type people were talking to
the State directly from Unit 2?" Ross stated.
"There was none that I'm aware of." (768)
Addendum 21
The first conversation on the incore thermocou-
ples involved Victor Stello and Mike Wilber. both
at IRACT. and Harold Kister at the regional
office:
STELLO (in background) : And then I'll
want to find out if they [can] give me a
core element temperature. I got the im-
pression those were not working. They
had thermocouples on all the outlet as-
semblies on the B&W plant. Do they have
any indication on thermocouples on the
assembly ?
WILBER : Harry '.
KISTER : Yes.
WILBER : We are talking about the fuel
assembly outlet temperatures. I under-
stand they do have thermocouples on the
fuel assembly outlet. Have they looked at
any of those?
KISTER : Are you thinking about West-
inghouse plant
WILBER: They are saying B&W has
that.
KISTER : B&W does?
WILBER : Yeah.
KISTER: Okay. Fuel element outlet
temperature right ?
WILBER: Yeah. (769)
Addendum 22
Characteristic of the flow of misinformation
concerning temperatures was a series of exchanges
that occurred between about 12 :15 and 12 :30 p.m.
on Wednesday. Donald Caphton and Eldon Brun-
ner at Region I were speaking with XRC inspec-
tor Walter Baunack. who was in Unit 1 at the
time. Baunack hypothesized that primary system
temperature was at saturation even though he had
no readings to go by to reach that conclusion :
CAPHTOX : How about "R" coolant tem-
perature, reactor coolant temperature,
Walt, anything on that ?
BACXACK: I suspect it's probably
pretty near saturated, wouldn't you
think, if they got a steam bubble in the
steam generator it would have to be
saturated.
BRUXXER : Xo reading?
BATTXACK: Xobody mentioned what it
was if that's what you are saying. (770)
XRC tape transcripts indicate that while the
above conversation was occurring. Donald Haver-
kamp. who was in the regional incident response
center with Caphton and Brunner was on another
phone speaking with James Higgins in Unit 2.
( 771 ) Minutes later. Region I's communicator with
IRACT provided headquarters with the following
update :
REGIOX I : They think the temperature
of the reactor coolant system has stabi-
lized. They feel it is saturated at 550 de-
grees fahrenheit.
IRACT: This is what is called a
hotleg?
REGIOX I: They say across the board.
IRACT: Isothermal ?
REGIOX I : That is what I'm telling you
right now. (772)
In this communication Region I reported that
liot and coldleg temperatures were both about
550F. when, in realitv. hotleg temperatures were
around 700 and coldleg temperatures some 450 C
lower than temperatures in the hotleg.
Circumstantial evidence from the tape tran-
scripts suggests two possible reasons for the above
misinformation being conveyed. One is that Hig-
159
gins erroneously reported to Haverkamp "T ave "
(average) readings for hotleg or primary system
temperature. That mistake had been made earlier
in the morning. Another possibility is that based
on Baunack's speculation that primary system tem-
perature was at saturation, Region I personnel de-
rived and reported a primary system temperature
which they had obtained by comparing the known
system pressure with standardized steam tables.
Either method would have produced the erroneous
information that system temperature was about
550F.
Addendum 23
The NRC did an analysis of HPI flow rates
based on changes in the level of water in the
Borated Water Storage Tank from which the HPI
water was drawn. (773) Its analysis showed that
over the four-hour period between 1 :15 p.m. and
5 :20 p.m., the average net rate of flow into the core
was 150 gallons per minute (gpm). During four
other periods of the day the rates were :
(1) 4 a.m. to about 7:30 a.m. (corre-
sponding to the period when the core was
first uncovered) 70 gpm ;
(2) 7:30 a.m. to about 11 a.m. (corre-
sponding to the period when the core was
again covered and repressurization oc-
curred) at least 680 gpm;
(3) about 11 a.m. to about 1:15 p.m.
(the period of the first depressurization)
360 gpm ; and,
(4) about 5 :23 p.m. to 6 :41 p.m. (the period
after the decision was made to repressurize
the system again) 470 gpm.
Addendum 24
Higgins had told the Special Inquiry Group:
Question : Do you recall any questions
or suggested questions coming in from
Region I or from Bethesda relating to
saturation conditions or relating to the
core being uncovered ?
HIGGIXS : No.
Question : Do you recall anybody over
the phone saying, "Hey, we think there's
a core coverage problem ?"
HIGGINS : Definitely not.
Question : You don't recall that ?
HIGGIXS: Definitely not, because there
were discussions among the caucuses that
went on as to Gary Miller saying the type
of thing : Does anyone here feel we're not
providing adequate core cooling or ade-
quate core coverage ? I didn't feel at that
time there was a problem. I didn't have
an indication the people on the other
end of the phone in Washington felt that
either. I guess I can add here things I
found out afterwards ?
Question: Sure.
HIGGIXS: Afterwards, that Mr. Stello
called the Unit 1 control room and talked
to an operator there sometime in the af-
ternoon and asked that, operator to pass
on to their management the NRO's con-
cern about core coverage, which if that
happened, it just never did get to the
caucuses, never did get to the right peo-
ple, and in fact, was really not the right
way to get it to management because,
first, coming from Mr. Stello at that
point, that's certainly a significant com-
ment because that represents some type
of NRC caucus, I would think, some type
of NRC consensus that had that feeling.
If I had heard that. I would have cer-
tainly taken some steps to find out why
they felt that and tried to communicate
that to Met Ed. (774)
Addendum 25
Another specific weakness in emergency plan-
ning was the lack of procedures for internal plant
communications during an accident, 7)articularly
with regard to diagnosing plant conditions. The
TMI-2 Emergencv Plan provided no guidance
to the emergency director about how to assess the
condition of the plant during an emergency if it
should be determined that the plant was in a state
that was not covered by the plant's emergency pro-
cedures. The plan merely delegated the responsi-
bility for developing internal plant communica-
tions procedures to the emergency director. Xor
were there any provisions in the Emergency Plan
for marshaling the technical and scientific advice
of outside agencies and groups. Tn fact, there was
no procedure in the Emergency Plan for participa-
tion by the reactor-vendor, the XRC or the archi-
tect's engineer in assessing plant conditions. (775)
160
Chapter 8
Recovery At Three Mile Island
161
con-
Cleanup workers at Three Mile Island gain access for the first time to the airlock leading to the ,...-
tainment Beyond the second door are the damaged reactor and other major components of the
plant s primary system
162
Chapter 8
Recovery At Three Mile Island
INTRODUCTION
One aftermath of the accident has been the
enormous and complex task of recovery. It can.
in fact, be considered a continuation of the
accident.
Recovery involves two phases: cleanup of the
TMI-2 facility, principally decontamination and
disposal of the radioactive debris, including the
damaged core : and the future disposition of Unit
2 whether to refurbish it as a power plant or to
decommission it.
The Special Investigation emphasized the clean-
up phase of recovery. Cleanup is of deep concern
to the utility, the XRC. the local population and
the Congress. It is unprecedented in scope and
complexity and is likely to have a substantial
impact on the future of nuclear energy in this
country.
The complex and uncertain steps involved in
cleanup are reviewed in this section not only in
terms of the technical difficulties, but also in terms
of financial, social, legal and regulatory consider-
ations.
The technical challenge is without parallel
among privately owned commercial nuclear power
plants. Coping successfully with the radioactive
debris, especially the core, is a very large and dif-
ficult part of the task. There are also health and
safety questions, such as the exposure of workers
to radiation and the proximity of TMI to a densely
populated area. However, based on the evidence
reviewed by the Special Investigation, including
prior recovery operations at government reactors
in the United States and other countries, the Sub-
committee believes that the technical challenge can
be met.
The technical questions are interwoven with fi-
nancial, social and legal factors. For example,
the potential cost to the licensee is great, and the
financial future of Metropolitan Edison (Met Ed) ,
the plant's operating utility, is unclear. Local
elected officials testified that the communities near
Three Mile Island are extremely apprehensive
about cleanup for many reasons, and very dis-
trustful of both the XRC and the utility. There
has been substantial opposition to many of the ini-
tial cleanup proposals. Some of the legal and
regulatory questions are without clear precedent.
The XRC has never dealt with a similar cleanup,
and it faces many unresolved issues. Especially im-
portant are the circumstances under which it may
take immediate action in authorizing cleanup
tasks, before the- required deliberative decision-
making procedures have been completed. Another
issue is the environmental review procedure to be
followed.
Within the context of these various issues,
cleanup poses a difficult dilemma. Cleanup requires
careful planning, but there is the pressure of the
unknown. The reactor's present condition is not
without risk, and the status of components vital
to the integrity of key systems is uncertain and
unpredictable. Further weakening and failure of
important equipment can be expected with the
passage of time, and accidental releases of radio-
activity and recriticality of the core are possible.
The various methods for venting the radioactive
gases, disposing of the radioactive water and re-
moving the radioactive waste all of which are re-
quired for cleanup could result in uncontrolled
releases to plant workers and surrounding commu-
nities. Even in cases where the XRC and GPU
have concluded that health hazards are minimal,
some members of the nearby communities view any
releases as hazardous. 1
To date, cleanup has followed established legal
and regulatory procedures that, while deliberate,
provide for orderly decisionmaking through care-
ful consideration of options and through public
participation. Decisions should involve the con-
sideration of timing and a careful weighing of the
risks and benefits of alternative courses of action.
1 See "Social Issues in Recovery," pp. 198, 199-200.
163
TECHNICAL ASPECTS OF RECOVERY
THE NATURE OF THE TASK
The accident at Three Mile Island badly dam-
aged the Unit 2 nuclear core and released radio-
activity into the primary system coolant water.
As of June 1980, the containment held hundreds
of thousands of gallons of the highly contaminated
water, whose lower layers have been described as
"flocculent in appearance, gelatinous, dark green
color," (1) the result of chemicals released when
the ftiel failed. The dominant radioactive isotope
in the water is cesium 137, 2 with a relatively long
half-life of 30 years.
The amount of water in the containment is still
increasing because of leaking pump seals. It may
threaten two motors which operate valves critical
to maintenance of the primary cooling system and
to removal of the radioactive water. Those valves
are also necessary for assuring the operation of the
new, long-term cooling equipment needed for
cleanup. Assuming plant conditions remain as they
were in April 1980, the valves should not become
submerged for at least a year.
The atmosphere in the containment consists of
various radioactive gases. Some accidental releases
have already occurred, and there is a possibility of
further ones. However, the potential amount is
slowly decreasing with time, as radioactive ma-
terial decays naturally.
Thus the atmosphere, walls and water in the
containment are all contaminated with radioac-
tivity. As of early June 1980. personnel were un-
able to enter the bnildin,<r to survey it. and estimates
of the levels have been based on indirect measure-
ments and analyses. The unsuccessful initial at-
tempt to enter the containment in mid-May raised
the question of whether corrosion would make de-
contamination more difficult. 3
Although actual levels of radiation are high,
thev are much lower than originally projected. In
July 1979, it was estimated that gamma radiation
would reach 2,400 rad/hr by December. (2) In
December it was calculated to be less than 1
rad/hr. (3)
There are two principal reasons for the substan-
tial differences between projected and actual levels.
The earlier estimates (4) had intentionally been
made very conservatively. In addition, they were
based on radiation levels as measured by the con-
tainment dome monitor. (5) Later independent
measurements 4 and tests of the dome monitor
showed that it was not functional.
Most of the isotopes with short half-lives al-
ready have decayed into their stable forms and no
longer emit radiation in the containment. This has
reduced the radiological hazard substantially.
Especially important is the decay of iodine 131
(1-131) , a volatile element that concentrates in the
thyroid gland. Because it has a half-life of about
eight days, virtually all of it has decayed. The
dominant radioactive isotope still present as of
June 1980 was krypton 85 (Kr-85), which has
a half-life of 10.7 years.
Radioactive water is also present in the auxiliary
building; again, cesium 137 is the dominant radio-
active isotope.
The damaged and highly radioactive core con-
tinues to generate relatively low levels of heat that
must be removed continuously.
The amount of radioactivity still present at the
site insubstantial and consists principally of long-
lived Isotopes. The condition and reliability of the
equipment and systems that must contain this radi-
ation are uncertain, particularly in the contain-
ment. The viability of critical electrical and
mechanical systems may have been affected by the
exposure to steam and moisture, the continuous
operation of equipment for much longer than de-
signed for, the cumulative effects of radiation, se-
vere thermal cycling 5 and the hydrogen burn that
occurred on the afternoon of the first day of the
accident.
Because the containment has been inaccessible,
the utility has been unable to evaluate the equip-
ment and systems directly. It has been impossible
to determine whether the elastomeric 6 seals and
the building air coolers 7 are undamaged.
SAFETY CONSIDERATIONS
As a result of the unknown and uncertain con-
dition of plant systems and equipment, the poten-
' Cesium 137 has been emitting most of the gamma radiation in the water. Gamma radiation is similar to. but more
penetrating than, X-rays. See "Radiation Effects and Monitoring," for a description of radiation and its measurement,
pp. 43-44.
s See p. 184.
1 A combination of techniques was used to make the measurements, including scans through a nine-inch hole bored
into an access port in the containment.
6 Thermal cycling the unusual hot and cold oscillating conditions during the accident may have altered the
mechanical strength of some of the system components.
This is a special pliable material used throughout the plant to prevent leaking.
The air coolers insure that air pressure inside the containment building is lower than outside in order that no out-
ward leakage will occur.
164
tial for some problems has been analyzed. These
include :
Leakage of krypton
If the elastomeric seals or air cooling units were
to fail, krypton would escape at ground level. I
could also escape if temperature in the containment
could not continue to be lowered sufficiently to keep
pressure down in order to offset the increase in
pressure that results from inward air leakage.
In response to questions by the Subcommittee,
then-NRC Chairman Joseph Hendrie 8 said he did
not think the krypton was a "pressing danger or
urgent risk." (6) However, he added, ". . . the
longer these materials are allowed to remain . . .
in the containment building, the more chance there
is that somebody will open the wrong valve or
something else will happen and some of it will get
out." (7) Because the krypton did pose "some in-
crement, however small it may be, to the public
risk." he concluded, "We need to get on with [re-
moving] it." (8)
Leakage of water through the containment walls
Results of a radiochemical analysis of the water
indicate that there is no short-term risk of corro-
sion attacking the %" steel lining of the building.
Nonetheless, localized concentrations of caustic
chemicals could produce leaks. It is not known how
long they might take to develop.
In April 1980 radioactivity was detected in one
of the eight wells drilled into the bedrock around
the plant in order to monitor the water continually
for leakage from the containment. The well was
located 60 feet from the Borated Water Storage
Tank. 9 A radiation level of 2,500 picocuries per
liter was measured, as compared with the normal
500 picocuries per liter attributable to natural
background radiation, 10 as measured prior to the
accident.
At the time the radioactivity was detected, the
GPU Service Corporation " did not know the
source of the contamination. Spokesmen said it
could have resulted from nothing more than fluc-
tuations in the natural background radiation level.
(9) It could also have been the result of leakage
from the Borated Water Storage Tank, since some
of the same radioisotopes were present in both the
tank and the well sample. (10) Another possibil-
ity was leakage from the primarv containment,
since, again, some of the same radioisotopes were
present both in the containment and in the well
sample. (11)
General Public Utilities Corporation (GPU)
considers the groundwater contamination to be im-
portant and has been investigating the source. (12)
If the containment building is leaking, little can
be done except to accelerate processing of the
water, although that action might involve some
risk and be undesirable.
Recriticality of the core, leading either to
limited melting of the fuel, or a core
meltdown
Both the NRC and the Special Investigation
looked into this possibility. (13) The NRC study
concluded that the most likely cause of recritical-
ity would be dilution of the boron concentration in
the water. 12 ( 14) According to the NRC, this proc-
ess would occur slowly enough that the approach to
criticality could be detected in time to take correc-
tive action, assuming the necessary instrumenta-
tion, procedures and equipment were available.
(15) The study further concluded that recriticality
most likely "[would] not result in significant off-
site radiological consequences." (16) Even in the
worst and least likely case a meltdown within an
unisolated containment with no means of removing
heat from the containment or no containment
sprays the latent risk of cancer to individuals off-
site would be negligible compared to the normal
incidence of that disease. (17) However, recriti-
cality could produce radiation levels in the con-
tainment at least 10 times higher than existed in
April 1980. (18) Those levels would constitute an
increased risk to the onsite workers involved in
cleanup. That risk would be reduced at the time of
recriticality by evacuating the workers, a protec-
tive action that is feasible because such an accident
would take place over 10 hours and radiation
would be released gradually. (19)
At its November hearings, the Subcommittee
asked Richard F. Wilson of General Public Utili-
ties Service Corporation. then-Direotor of Cleanup
at TMI-2, about the possibility of core melting.
He testified,
I don't believe [that with] the current
heat production in the core, there's any
credible set of circumstances which would
lead to melting of the core. (20)
Several NRC officials also testified. Harold R.
Denton, the Director of the Office of Nuclear Re-
actor Regulation (NRR), said ". . . there's no
possibility that there would be a core melt through
the reactor vessel. . . ." (21) However, he raised
8 Hendrie was replaced by Acting Chairman John F. Ahearne on December 7, 1979.
' See "How the Plant Works." p. 31.
10 See "Radiation Effects and Monitoring," p. 45.
11 GPU Service Corporation performed engineering functions for GPU, such as TMI-2 cleanup, etc. Recently, GPU
formed a new entity, GPU Nuclear Corporation, in part to improve coordination among the member utilities.
12 See "How the Plant Works," p. 29.
165
another possibility that hot spots 13 would de-
velop within the core, leading to localized melting :
. . . I'm not quite so sanguine about
whether or not . . . temperatures might
not approach melting somewhere in the
fuel rods themselves. . . . (22)
Since the core's configuration is badly distorted,
some areas may not be getting cooled. Richard
H. Vollmer, then-Director of TMI Site Support
and Assistant Director for Systems and Projects,
Office of Nuclear Reactor Regulation, NRC said
that the consequences of the hot spots would be
minimal, as most of the fission products in the core
that have high volatility had already been released
from the system or had decayed. Thus, according
to Vollmer, ". . . even if a small portion of the
core were to attain high temperatures, it would
not pose the usual threat to the public health and
safety." (23) He went on to say, "It would . . .
basically [involve] solid fission products which, if
released from the core, would likely condense [in]
the primary system or containment and not pose
an outside threat." (24)
As noted, dilution of the boron concentration
could cause recriticality. Vollmer testified that the
boron concentration could accidentally decrease as
a result of "boron precipitation, which usually oc-
curs on the colder portions of the surfaces in the
primary system. . . ." (25) However, he said it
was unlikely that the precipitation would occur
near the core, since it was the hottest part of the
system. (26) He also noted that because of temper-
ature considerations, "It [boron] should be ex-
pected to stay in the solution." (27) Chairman
Hendrie likewise said, "I am not very concerned
about losing boron out of that reactor water." (28)
However, their analyses did not describe diffi-
culties in detecting decreases in boron concentra-
tion, as pointed out in an NRC memorandum. (29)
First, as of April 1980, there was onlv one opera-
tional neutron detector. If it were to fail, monitor-
ing anv increase in power, a signal of recriticality,
would be severely hampered. Second, an unforeseen
rapid decrease in boron concentration would re-
cniire equally rapid operator action in turning on
the decav heat removal pumps. C30) The NRC
memorandum pointed out a third factor. As of
April, the boron concentration was being measured
at a point 200 feet from the core. The amount at
that point would not necessarily reflect the concen-
tration in the core.
Inability to remove decay heat
If circulation should stop, so that decay heat is <
not being removed, and assuming the water level
is maintained, the temperature of the coolant
water would not reach the boiling point so long as i
adequate pressure is maintained in the system. (31)
The core would heat the primary coolant until a
balance is established at the point where natural
heat losses " equal the heat generated within the
reactor. If this condition is reached, temperature
in the reactor would remain steady. At the level
of decay heat in November, this balance would oc-
cur when the temperature of the system reached
500-600 F (260-315 C). (32) Since fuel melt-
ing occurs at greater than 5.000 F (2.760 C), the
likelihood of fuel melting from this sequence of
events is extremely low.
INCIDENTS SINCE THE ACCIDENT
There have- been several problems at the plant
since the accident.
On December 21, 1979, the NRC issued a pre-
liminary notification that small amounts of kryp-
ton 85 were being released from the main
condenser; they were detected by the condenser
radiation monitor. 15 The amounts were measured
at 0.0002 microcuries 16 (two-tenths of a billionth
of a curie) per milliliter and represented no health
hazard." (33)
GPU explained that because pressure was higher
in the containment than in the secondary system
piping, it believed the gas was leaking from the
containment into the steam line of the steam gen-
erator "A" through degraded valve seals. The gas
was then picked up by the steam traveling through
the line from the generator to the condenser.
On February 11, 1980. coolant was released from
the primary system during surveillance testing of
the make-up pumps. By the time the leak was veri-
fied and stopped, between 600 and 1.000 gallons
had drained onto the floor of the make-up pump
cubicle in the auxiliary building. The water con-
tained a small concentration of dissolved krypton
85 in addition to 50-100 microcuries per milliliter
of cesium 137. (34) It gave off krypton 85 gas, and
an estimated 200-300 millicuries (two-tenths of a
curie") were released to the atmosphere through the
ventilation system exhaust of the auxiliary build-
ing. (35) However, the station exhaust vent radi-
ation monitor registered no increase in radiation.
This means that localized melting might occur, but that progression to a full meltdown is unlikely.
"Natural heat losses from the reactor occur through conduction, convection and radiation. An analogy is a light
bulb. The bulb heats up until its temperature reaches a point where the heat losses equal the heat generation all
without the aid of a fan or cooling water. It reaches this temperature and stays at that temperature until the light is
switched off.
10 The condenser is designed to release some radioactive gas during normal operation. The amount is regulated,
and anything over it will be picked up by a radiation monitor.
""The curie is a unit of measurement that describes the amount of radiation present or released (see Technical
glossary). One microcurie is one millionth of a curie. One picocurie is one millionth of one millionth of a curie.
" See "Radiation Effects and Monitoring," p. 45.
166
I iin (I the water itself drained into a sump from
,, which it was routed into one of the auxiliary build-
ing tanks.
^veral of the 12 workers who entered the cubi-
* clc to locate and stop the leak were exposed to the
radiation. The maximum individual whole body
i dose received was about 160 millirem, (36) within
the limits set by the NRC.
On February 12 and 13, there was an additional
I release of krypton gas. Approximately 4 curies
I were cm it t cd to the atmosphere while workers were
collecting a sample of air from the containment.
The \KC concluded that the leak on February
11 \\a- the result of equipment failure. (37) I>in'-
in<_' a briefing February !">. Victor Stello, Jr.. J>i
r of the Oflicc of I n -|ie<-t ion and F^nforcement
I A' ]}), exr)Iained that a discharge pressure in-
strument line valve lxcame dislodged when one of
the make-up Dumps was restarted. (38) An NRC
inquiry found that the releases on February 12 and
]'.', occurred because "The shift engineer who im-
plemented the air sample procedure did not use the
effective Dmcedure . . . because the individual did
not obtain a controlled copy of the procedure. ..."
('.','n The inquiry concluded that the failure to fol-
low document control procedures would be subject
to subsequent enforcement action. (40)
These incidents and the continued uncertain
status of important equipment in the plant high-
light the need for prompt attention to planning
Cleanup and for improvement in the utility's radi-
ological monitoring and protection program.
THE STEPS IN RECOVERY
There is still no carefully structured, overall
plan for the cleanup at TMT, in part because of
technical, regulatory, legal and financial uncertain-
ompoundcd by the inability of the utility to
enter the containment to conduct a detailed evalu-
ation. Thus the specific steps are still undefined.
Generally, recovery will take place in two
phases:
Cleanup, which involves
maintaining plant stability and pre-
venting releases of radiation ;
decontaminating the plant and dis-
posing of waste materials ; and
Deciding on the future of the facility.
These are summarized below briefly and are dis-
d in greater detail later in this section.
MAINTAINING STABILITY
Reactor stability must be maintained while pre-
paring for and conducting cleanup. Two factors
-ential in controlling the reactor core : assur-
ing -ulK-riticality and continued cooling.
Because some control rods are believed to have
melted, and the shape of the core is distorted, sub-
criticality can only be accomplished by maintain-
ing a sufficiently high concentration of boron in
the coolant. (41) Proper cooling requires that the
core remain covered with coolant. Further, the de-
cay heat being generated by the core imr-t be re-
moved continuously, whicn is accomplished by
circulating the coolant around the core. In addi-
tion, the coolant must be kept from boiling, which
requires keeping pressure in the primary system at
a certain level. Hence, a functional system for con-
trolling pressure is important, to stability. Finally,
the primary coolant system (including the reactor
vessel, piping and pressurizer) must remain intact
in order to maintain the needed water level. A large
break in the piping, for example, would lead to a
release of coolant and thereby threaten reactor
stability.
If all coolant were lost, and the coi-e remained
uncovered, it would gradually heat, but not melt.
In an interview with the Special Investigation
staff, Richard H. Vollmer, Office of Nuclear Re-
actor Regulation, NRC. explained:
If you lost all the water, ... in the pri-
mary system. I think even then, [the li-
censee] would be able to take the heat out
because the core would go up to elevated
temperature where it would start to radi-
ate to the vessel . . . and you would have
your conduction that way. (42)
As noted, the rising level of water in the con-
tainment could eventually submerge and cause
failure of two valves that are important to the new
long-term cooling equipment to be installed. This
poses a threat to stability.
CLEANUP
The first step in cleaning up the plant is to
decontaminate the auxiliary building. Prior to
decontamination, a buildup of radioactive water
in the auxiliary building caused the greatest short-
term possibility of a release of radiation to the
environment. (43) As a stopgap measure, the
water has been stored in tanks, but ultimately it
must be processed, both because the radioactivity
must be removed and because the capacity of the
tanks is limited. Furthermore, much of the equip-
ment necessary for controlling the stability of the
plant is in the auxiliary building. If this equip-
ment is left near the highly radioactive tanks, the
workers who must operate it will receive unneces-
sarily high dose rates of radiation and would have
to be replaced frequently by other workers.
The next major step in cleanup is to remove the
krypton 85 gas inside the containment to provide
safe access to equipment inside the building, such
as the reactor and steam generators, and to permit
decontamination.
167
Next is removal of the highly radioactive water
inside the building. This water represents a health
hazard to workers who will have to spend long
periods inside the building and also impedes over-
all decontamination.
After removal, the water will be processed to rid
it of radioactive debris, using filtering equipment
similar to that used in the auxiliary building (see
pp. 184-185).
To gain access to the core, the next step in clean-
up, the reactor head must be removed. Workers
must first disconnect components such as the con-
trol rods that run through the head and into the
core. This job will be difficult if these components
are jammed or entangled as a result of the distor-
tion of the core. However, techniques have been
successfully applied to similar tasks in previous
recovery efforts. (44)
Next is removal of the core. 18 Removal of the
head and core is difficult to plan at this stage be-
cause the specific damage is not known. The actual
steps will be based on visual examination and
mechanical tests that cannot be performed until
there is unrestricted access to the containment.
Although cleanup is technically challenging and
represents a hazard to workers, comparable tasks
have been carried out successfully at other plants
severely damaged by accidents. 19
Much of the technology to be used in cleanup is
based on that developed principally at govern-
ment-owned facilities for other applications,
including previous accidents. In the previous ac-
cidents, however, cleanup could be accomplished
relatively quickly and at minimum cost because
government plants were typically self-sufficient
complexes where administrative support, person-
nel and disposal sites were readily available, as was
decontamination technology. In addition, prior
accidents in the United States involving releases
predated the deliberative requirements of the Na-
tional Environmental Policy Act of 1969 involv-
ing environmental impact statements and direct
public participation, as applicable to the area of
atomic energy.
The technology already developed in these prior
cleanup exercises is being made available for the
Three Mile Island cleanup. (45)
Two matters that will have to be addressed in
relation to cleanup are the disposal of radioactive
wastes and the worker-safety program. Both mat-
ters are also discussed in detail below.
No matter what is ultimately done with the
plant, the cleanup must be completed. The plant
is now unsafe in comparison with a normal reactor,
and the likelihood of further accidents ac-
cumulates with time.
FUTURE DISPOSITION OF TMI-2
Once cleaned up, TMI-2 may be decommissioned
(taken out of service permanently) or rebuilt
either as a nuclear or as a coal-fired facility.
A decision on the plant's future cannot be made
now. Its overall physical condition must be better
understood, and there are financial, social, legal
and regulatory issues that must be resolved.
COST AND SCHEDULE
Soon after the accident, General Public Utilities
Service Corporation (GPU Service Corpora-
tion) 20 hired the Bechtel Power Corporation to
perform an analysis of the recovery of Unit 2. (46)
The study, begun shortly after the accident, in-
volves three phases. The first, completed in July
1979, outlined a technical plan for cleanup through
the stage of building decontamination (excluding
core removal) and estimated the costs of recovery
through recommissioning. The study was neces-
sarily based on very preliminary information and
involved best guesses in many cases.
Bechtel estimated the cost of recovery, including
refurbishment of the TMI-2 plant and replace-
ment of the core, to be about $400 million. 21 exclud-
ing energy replacement 22 and certain other costs.
It assumed a period of 4 years for the entire
recovery from the time workers first entered the
containment, 23 and a manpower requirement of
approximately 4.1 million man-hours. Cleanup
alone would involve more than 1,000 persons at
any one time, to be drawn from the national pool
of radiation workers. Decontamination would
necessitate large amounts of protective clothing
and equipment. For example, an estimated 1 mil-
lion each of plastic coveralls and hoods, breathing
cannisters and rubber gloves would be needed.
For cleanup alone, Bechtel projected a schedule
of about 2 years and a figure of about $200 mil-
"The half-life of the fissionable uranium 235 (U-235) in the reactor is 713 million years. This explains in part
the need to remove the core. This task will be one of the last and most difficult.
" Serious reactor accidents that have been cleaned up include : the SL-1 facility in Idaho ; two at the Chalk River
facility in Canada ; Enrico Fermi in Michigan ; Windscale in England : and the SRE in California. These accidents and
aspects of their cleanup are described in "TMI in Perspective : Other Nuclear Accidents," Appendix A, pp. 221-226.
M See "Prior to the Accident," p. 51, for details on GPU Service Corporation.
21 In early June, GPU indicated that final costs of cleanup and refurbishment could far exceed its initial $400 million
estimate. See "Financial Aspects of Recovery," p. 190, and see fn. 86, p. 191.
M Purchase of energy from other utilities to supply customers of the GPU system.
n Since workers still have not entered the containment, the 4-year period has not yet begun.
168
lion. 24 It should be noted that this figure did not
include the costs of in-service inspection (to re-
qualify undamaged equipment), reconstruction,
refurbishing of major equipment, radioactive
waste disposal and miscellaneous cleanup tasks.
Xor did it include the expense of replacing the
core, estimated to be between $60 and $80 million.
An independent study of the costs of recovery
performed for the President's Commission on the
Accident at Three Mile Island estimated decon-
tamination and fuel removal at $90-$130 million.
(47)
The final cleanup figure may vary substantially
from the estimates of both Bechtel and the Presi-
dent's Commission. Since the plant must be cleaned
up. the associated costs for that portion of recovery
are unavoidable.
The second phase of the Bechtel study, sched-
uled for completion soon, is to cover removal of
the reactor head and disposal of the core. The third
and final phase will address recertification and
recommissioning the steps necessary to put the
plant back into operation. The studv of this phase
is incomplete, since it requires a detailed assess-
ment of the plant, which cannot be finished until
the containment is entered.
PLANNING FOR CLEANUP
All planning for cleanup and recovery must be
coordinated with the XKC. The agency must ap-
prove the cleanup plan and is responsible for es-
tablishing the requirements governing cleanup,
including limits on worker exposure and radiation
releases. Because this is the first major commercial
accident in the United States 25 involving large-
scale cleanup and recovery, mechanisms for coor-
dination between the XRC and the utility are being
developed as recovery proceeds.
As of June 1980. the XRC had not approved an
overall plan for cleanup. GPU Service Corpora-
tion prepared a plan and schedule in response to a
subcommittee request. The XRC reviewed the plan
and provided three alternate schedules, reflecting
three contingencies.
With no approved plan, there can. of course, be
no final target schedule or timetable for comple-
tion of the cleanup. The XRC's most conservative
estimates of time for the cleanup run nearly 5
vears. (48) morp than twice the Bechtel projection.
The lengthy XRC schedule allows time for en-
vironmental reviews and the application of more
stringent restrictions on radiation releases.
The XRC also had not, as of March 1980, issued
interim guidelines on acceptable releases from the
plant, Xormally a nuclear power facility is al-
lowed, and does, release a specified, non-hazardous
amount of radiation per month. Xonetheless, a
"zero-release" standard had effectively been im-
posed by the XRC, which made any action on
cleanup difficult.
The lack of an XRC-approved plan and schedule
for cleanup and of new NRC requirements gov-
erning the work became major issues. At the Xo-
vember 8, 1979, Subcommittee hearing, the Special
Investigation staff reported that "more than 7
months have elapsed since the day of the accident,
but there is still no overall plan for recovery." (49)
The subcommittee chairman stated :
Our preliminary findings indicate that
the Nuclear Regulatory Commission ap-
pears to be withholding guidelines for
such a plan until the utility makes its pro-
posal, while the utility position is that
such a plan cannot be developed until spe-
cific regulatory guidelines are provided
by the XRC. So we now seem to find our-
selves in a situation where the XRC and
Metropolitan Edison are each waiting for
the other to make the first move.*' (51)
As noted, a vear after the accident, there were
four possibilities : three schedules developed by the
XRC, and one unapproved plan developed by the
utility.
With no plan and no guidelines, no major prog-
ress had been made toward full-scale cleanup as
of early June 1980. Using that date as a starting
point and taking the most optimistic timetable, re-
moval of the core a principal health and safety
concern would not occur until sometime in 1982.
According to GPU Service Corporation, in order
to meet even that target and other cleanup sched-
ules, portions of its plan should have been set in
motion early in 1980. (52)
On Xovember 9. 1979, the Subcommittee re-
quested that the XRC supply a best-estimate plan
bv December 20, 1979. to include a timetable for
the entire cleanup and. in addition, its plans for
coordination with GPU. The XRC submitted to
the Subcommittee material that did not include a
best-estimate plan, but instead outlined four pos-
This includes a 33 percent contingency. The contingency allows for the preliminary nature of the facts upon
which estimates of cleanup were based, the potential for pricing changes and an assessment of productivity The report
maintains that productivity is a variable which depends on conditions in the containment, administrative controls
required support, worker dose limits and. finally, availability of special materials, equipment and many other items.
["here was also an accident at the Enrico Fermi reactor, a small commercial, power-producing "fast" reactor
imilar to TMI-2. The Occident was contained within the primary system and hence was not as serious as that
Three Mile Island in Perspective : Other Nuclear Accidents." Appendix A. p. 225, for further discussion
The Subcommittee Chairman's statement was based upon two internal memoranda generated bv the Special
Investigation staff (50).
169
5U-OS8 0-80-12
sible cases with widely differing schedules ranging
from 38 to 58 months. The Commission explained :
Because of the impact on schedule that
could result from environmental reviews
and subsequent equipment and opera-
tional restrictions, four decontamination
program cases were compared to bound
the likely duration of the decontamina-
tion process. (53)
The schedules provided by the NRC did not
allow time for public hearings :
It was also assumed that no hearings
would be held for any steps during Phase
1 and Phase 2 [the cleanup phase of re-
covery]. (54)
In a discussion between John Ahearne, the new
Chairman of the NRC, and Richard H. Vollmer
during a meeting of the Nuclear Regulatory Com-
mission on November 29, the mechanics of the en-
vironmental review process were outlined :
AHEARNE : Do you have embedded any-
where in there the concept of will there
be any hearings ?
VOLLMER : Hearings were not really em-
bedded in here and as I indicated our esti-
mates of the environmental assessment
and perhaps, more particularly, the en-
vironmental impact statement are prob-
ably as skinny as they could get. (55)
Regarding guidelines on releases, a point also
raised by the Subcommittee, the NRC said, in its
response :
We intend to solicit public comment,
within the context of the draft program-
matic environmental impact statement for
the TMI decontamination and cleanup
activities, on whether these limits, which
were developed for effluents resulting
from normal operations, are appropriate
for the TMI cleanup activities in light of
the differences in the volume and duration
of the release of such effluents. (56)
Finally,
The staff anticipates that existing Com-
mission regulations, guidelines and cri-
teria applicable to a normally operating
facility, will continue to be applied to
cleanup activities at TMI-2. However, we
recognize that although certain activities
would otherwise be permitted at a nor-
mally operating facility, it may be war-
ranted, in the public interest, to prohibit
them at TMI-2 even though they could
be conducted in full compliance with ex-
isting effluent limitations in the operat-
ing license or NRC regulations, until
further evaluation of them is under-
taken. (57)
In its response to the Subcommittee, the NRC
made reference to two other key issues that have
a bearing on adoption of a plan and determination
of a schedule.
First is the reference to a programmatic environ-
mental impact statement. On November 21, 1979,
the NRC decided to prepare such a statement, a
task requiring at least a year. (It is scheduled for
completion in September 1980.) If no "emer-
gency"-type situations occur in the interim, it is
likely that all decisions on cleanup will be deferred
until then."
The second reference was to the possibility of
more stringent requirements being placed on TMI-
2, for example, in connection with releases. Her-
man Dieckamp, President of GPU, had testified
on this point on November 8, 1979 :
If we were to be able to proceed on the
basis of existing regulations and specifi-
cations, one would be able to proceed to
discharge some of the water which was
contaminated in the accident after having
been processed. But the whole process,
institutional process, has, in effect, frus-
trated that. (58)
The NRC Task Force Report
Early in February 1980, two small, uncontrolled
releases of radiation occurred at the site, as noted.
During the week of February 11, Commissioner
Gilinsky sent Victor Stello, Director of the Office
of Inspection and Enforcement (I&E), to the site
to assess the situation. As a result of Stello's visit,
a task force on the cleanup at Three Mile Island
was established on February 15, 1980, under the
direction of NRC's William J. Dircks, Acting Ex-
ecutive Director for Operations. He directed the
task force to complete a report for the Commission
by February 29, 1980, that would "evaluate the
cleanup operations at Three Mile Island, how they
are being accomplished, and the rate at which they
are being accomplished to insure that the public
health and safety is being protected." (59)
Selected findings of the Task Force (60) were
that :
The maintenance of TMI-2 in a stable
condition cannot be accomplished with
zero radiation releases.
The November 21, 1979. Policy State-
ment of the Commission is being inter-
preted by the NRC staff as a "zero re-
lease" requirement insofar as it affects
cleanup.
Both NRC and the licensee have al-
lowed what was once a relatively high
=1 See pp. 201, 204-207 for further details on the Programmatic Environmental Impact Statement.
170
priority on developing and implement-
ing TMI-2 cleanup plans to erode. . . .
The full extent of approval authority
of the XRC TMI Support Staff is un-
clear
* * *
The Commission's Policy Statement
provides sufficient flexibility so that
prompt actions which are shown to be in
the best interest of the public health and
safety may be undertaken by the Com-
mission prior to completion of the PEIS
[Programmatic Environmental Impact
Statement] ... If such prompt actions
become numerous and must go to the Com-
mission for approval, delays will be intro-
duced. . . .
Neither the XRC staff nor the licensee
lias proposed a set of criteria that would
provide an interim envelope for the con-
duct of day-to-day activities . . . pending
completion of the PEIS. . . .
* * *
The completion of the PEIS has be-
come an important milestone in the clean-
up of TMI-2. However, the Commission's
intended use of the PEIS after comple-
tion is not clear to the XRC staff. . . .
* * *
Xeither XRC nor the licensee has given
sufficient consideration to concerns re-
lated to the waste form for ultimate dis-
posal of TMI-2 waste off-site. . . .
The recommendations of the Task Force were
that the :
Commission announce a commitment to
proceed with the cleanup of TMI-2 in
as expeditious a manner as possible.
Schedules for staff and licensee actions
should be established and closely moni-
tored bv EDO [XRC's Executive Direc-
tor of Operations]. . . .
* * *
Commission establish and EDO en-
force priority system that places clean-
up and PEIS preparation higher than
issuing new operating licenses. . . .
* * *
EDO ensure cleanup has adequate re-
view for long-term waste impacts by hav-
ing full staff coordination on all waste
disposal actions. . . .
Staff immediately propose for Commis-
sion approval rational, conservative in-
terim criteria to permit releases asso-
ciated with plant maintenance and
data-gathering for future cleanup re-
quirements while awaiting completion of
PEIS. An environmental assessment
would be prepared for establishment of
of these criteria, and CEQ would be con-
sulted. The need to provide opportunity
for public comment should be consid-
ered.
On April 7, 1980 the Commission approved a set
of interim radiological effluent criteria, allowing
some work to proceed. (61)
Pennsylvania Governor's Commission
On February 26. 1980, Governor Richard Thorn-
burgh of Pennsylvania issued the report of the
Special Governor's Commission on Three Mile Is-
land. (62) One of the topics covered was cleanup.
The Commission cited the various risks present at
the plant and raised several of the major issues:
. . . decontamination of the water stored
in these [the auxiliary building] tanks is
essential. . . .
* * *
The major advantage of the controlled
[krypton] venting option is that it can be
accomplished in a relatively short period
of time and it is a permanent disposal so-
lution. The alternative disposal system*
create large volume* of intensely concen-
trated waste material which must be
stored on-site or transported to a perma-
nent disposal facility. These are not per-
manent solutions, and would continue to
impose a potential public health hazard.
[Emphasis in original]
* * *
[Limited access to low-level radioactive
waste repositories in South Carolina and
Washington State] may evolve into a se-
vere problem for Pennsylvania.
During the week of April 7. 1980, the Union of
Concerned Scientists agreed to a request by Gov-
ernor Thornburgh to perform an independent
study of krypton venting. The XRC was not re-
quired to await the outcome of the study but did so.
DOE and EPA Involvement
The Department of Energy (DOE) concluded,
independently, that controlled purging was the
preferred alternative for removing the krvpton.
On February 5. 1980, G. W. Cunningham, Assist-
ant Secretary for Xuclear Energy. DOE, sent a
letter to Dircks, which stated :
The purpose of this letter is to urge the
Commission to act promptly on the mat-
ter [of krypton venting], . . . (63)
The Environmental Protection Agency (EPA)
is also involved in the cleanup. Herbert Feinroth
of DOE explained that :
. . . shortly after the accident, the White
House asked the Environmental Protec-
171
tion Agency to coordinate the roles of the
several agencies, including DOE, NRC,
and the State in a long-term environmen-
tal monitoring plan on [TMI]. . . .
They [EPA] published a long-term
surveillance plan which they have been
conducting in the last year. This past
week, they have initiated an activity to
update that plan to include specifically
what extra things should be done should
the Commission approve the venting pro-
posal. (64)
RADIOACTIVE WASTE DISPOSAL
Storage, shipment and ultimate disposal of
radioactive wastes produced and accumulated
during cleanup present additional problems, as
does disposal of the highly radioactive core. There
are only three available commercial disposal sites
for low-level wastes (one each in Nevada, South
Carolina and Washington), and none for high-
level wastes. 28 Both Nevada and South Carolina
have requested that TMI-2 wastes not be sent
there, and all three States have a reciprocity agree-
ment that bars disposal in all three should a ship-
ment be found in violation of the requirements of
any one of the States. (65)
In the State of Washington, nuclear waste be-
came a major 1980 campaign issue. Governor Dixy
Lee Kay has said that no out-of-state low-level
wastes should be allowed in after December 31,
1982, a three-year period that was to allow States
time to develop other options. (66) In addition,
the Governor of South Carolina has imposed grad-
uated limits on the amounts of low-level wastes
that will be permitted into that State, according to
testimony before the Nuclear Regulation Subcom-
mittee on January 25, 1980.
Taken together, all three provisos raise serious
questions about the availability of disposal sites
for the substantial quantities of low-level wastes
that will be generated over the next several years
as part of the TMI cleanup.
In the case of high-level waste, the problems are
compounded. Because commercial reprocessing of
spent fuel has been deferred indefinitely in the
United States, 29 the Unit 2 fuel may have to be
disposed of unaltered. However, no commercial
disposal sites are available for high-level trans-
uranic wastes. In remarks before the Subcofnmit-
tee, NRC's Denton said :
Some of the waste will be high-level
waste, as opposed to low-level waste. And
I'm sure you're aware there's considerable
difficulty in the country today with dis-
posal of low-level waste. It's not clear to
me that the depositories for high-level
waste will be available in the time frame
of cleanup, and it may become necessary
that some of these [will have to] be stored
onsite until that issue is resolved. (67)
At present, low-level waste at TMI is being
handled in two ways. First, interim facilities have
been constructed onsite to meet the immediate need
for the storage of those low-level wastes extracted
in the processing of radioactive water with
EPICOR. However, the capacity will not be suffi-
cient to accommodate all the anticipated wastes
from cleanup. Moreover, the TMI site cannot be
used for long-term storage, since it fails to meet
requirements as to the depth of the water table and
geological characteristics necessary to assure that
any accidental leakage of radioactive material will
not spread in the environment.
Second, some waste is being disposed of offsite.
The Nuclear Engineering Co. (NECO) , which op-
erates disposal sites in Richland, Washington, and
Beatty, Nevada, is accepting the TMI-2 wastes at
its Richland facility. TMI also has a contract with
the Chem-Nuclear Company to transport, handle
and deliver wastes generated by decontamination
of the auxiliary, fuel handling and diesel genera-
tor buildings and other areas as "requested by
Met Ed."
How many truck shipments will ultimately be
required is uncertain. During the Subcommittee's
November 8 hearing, Wilson of GPU Service Cor-
poration said :
I don't think we have the total number of
that [shipments], because to some degree
it depends what form it [waste] will even-
tually be removed from the site, but cer-
tainly will amount to be in the many,
many hundreds of shipments. (68)
The Bechtel study had made a rough estimate
of 2,440 truckloads (see Table 1). (69)
! * High-level waste is legally defined in 10 CFR 50 Appendix F. Generally speaking, it is that concentrated liquid or
dried waste obtained from the first cycle of extraction during the reprocessing of irradiated reactor fuels. Low-level
waste is everything that is not high-level.
28 Reprocessing is a technology in which spent fuel is chemically processed to remove plutonium and depleted
uranium for recycling as fresh fuel. President Carter has placed a moratorium on commercial reprocessing. Conse-
quently, spent fuel Is now stored at the various sites in spent fuel pools.
172
TABLE 1. Estimated number of radioactive waste ship-
ments for different categories
-V umber of
standard tf eight
Type of tcatte truck shipments
High-level resins 37
Intermediate level i to 7 drums at 10 curies per
drum, per truckload 500
Low level 15 to 18 drums at less than 1 curie per
drum, per truckload 300
Dry compacted low-level wastes 50 to 60 drums per
truckload 1-250
Major components (air coolers, pump motors, etc.)
(approximately) _
Total shipments 2- 440
This table was developed based on the assump-
tion that the krypton 85 and processed water
would not need snipping and does not include
shipment of the core.
Because of its limited volume, the water from
the auxiliary building can be stored and disposed
of with only minor difficulty. However, when the
entire cleanup task is considered, the volume be-
comes substantial. For example, about 1 million
gallons are in the auxiliary and containment build-
ings combined 250.000 and 700,000 gallons re-
spectively. 30 In addition, the utilitv estimates
another 7 to 9 million gallons will be used for
cleanup. Of this amount, up to about 5 million
gallons of processed water will have to be disposed
of. The rest will be purified and reused during
cleanup.
By May of 1980, the water in the auxiliary
building tanks was being processed by EPICOR-
II at an average rate of 3 gallons per minute. This
system operates much like a home water softener.
It contains a series of demineralizing filters, each
of which in succession removes radioisotopes from
the water and traps them on a resin bed. The bed
is attached to a liner that is removed and stored
once saturated.
The processed water is being held in clean water
storage tanks, in part because of public opposition
to its discharge into the Susquehanna 1 River. 31
Other options being considered for its disposal
include reuse in other decontamination tasks on-
site. evaporation or solidification in cement.
Waste solidification will have its first TMI-2
application in connection with the saturated resin
beds of the EPICOR-II system. When these are
changed, the water is drained from the system, but
a small amount remains behind. Ordinarily, this
residual water is partially removed prior to ship-
ment by vacuum de-watering techniques. 32
The States which regulate commercial radio-
active waste disposal sites have set a standard that
less than 1 percent of the volume of the radio-
active waste received can be water. While this
standard can be met with vacuum de-watering,
errors in de-watering sometimes result in greater
than acceptable percentages of water. Conse-
quently, as an additional step toward safe handling
and shipment of these TMI wastes, the XRC has
ordered that the resins from EPICOR-II be solid-
ified in cement to immobilize the radioactive waste
completely. (70)
The licensee plans to install cement solidification
equipment sometime in 1980 and will put the
wastes into 55-gallon drums. (71) These can either
be stored in underground silos onsite or shipped
offsite for burial.
Nonetheless, the NHC maintains that the licensee
has, or can construct, enough radioactive waste
storage areas onsite to preclude the need for ship-
ping the solidified EPICOR-II waste offsite. (72)
As noted, one commercial low-level waste site
was still available in June 1980, and disposal of
low-level wastes should not pose a problem in the
near future. Further, despite concerns expressed
by the communities surrounding TMI (73), the
XRC has said that onsite storage of some wastes
from the cleanup will not pose any greater health
haiard to the workers, the public or the environ-
ment than it does at other commercial plants. (74)
The longer term storage of TMI wastes is, how-
ever, a problem. The XRC has not yet dealt di-
rectly with the possibility of the closure of the
remaining low-level waste disposal sites, although
it has asked DOE to prepare a contingency plan
for that event. (75)
The issue of closure goes beyond TMI. Perma-
nent closure after 1980 would directly affect all
nuclear power plants. Onsite storage would quickly
become exhausted, and eventually the plants would
have to close down.
With respect to the core, while the Federal Gov-
ernment is seeking solutions to long-term storage
(e.g., through DOE). TMI may reauire a nearer
term solution. GPU has asked for DOE's assist-
ance in determining how the reactor fuel can be
transported and what can be done with it, since
that Department is presently responsible for find-
ing solutions to the overall nuclear waste problem.
DOE is studying various options for storing the
TMI core, to' be completed by July 1980. 33 Poten-
" These figures reflect the volume in April 1980.
11 See "Legal and Regulatory Aspects of Cleanup." pp. 201-204. 207.
* Vacuum de-watering is accomplished by lowering the pressure to a point where the water will naturally boil
off. The resulting vapor is condensed into non-radioactive water, leaving behind the radioactive resin beds.
a The Subcommittee staff was informed by Herbert Feinroth. Chief. Nuclear Reactor Evaluations Branch, Divi-
sion of Nuclear Power Development, DOE. that his division is funding the study and that only the technical feasibility
of storage options is under review.
173
Storage vaults for radioactive wastes resulting from cleanup
174
tial interim technical solutions being considered
include :
Storage in fuel pools onsite or elsewhere;
Storage in shielded facilities;
Eeprocessing.
An additional problem affecting interim fuel
storage is the abnormal chemical state of the de-
bris, which consists of a combination of zirconium,
hydrides, oxides, fission products and possibly
various alloys and other chemical compounds. Its
corrosive characteristics must be determined be-
fore the long-term integrity of storage containers
can be assured. When asked by the Subcommittee
about these problems. Denton said :
I think they [core materials] will present
some very interesting technical questions.
In what "form . . . these high-level wastes
should be solidified. How should this
really high-level waste be contained . . .
you heed to know . . . what type of envi-
ronment are they expected to be in over
their lifetime, in order to put them in
proper form to begin with. (76)
Commissioner Hendrie also commented on the
storage options for the core :
. . . there are about two options in the time
frame we are talking about. One of them
is to keep the casks onsite for some pe-
riod . . . and the other one would be for
one of the major Government processing
centers to accept those casks . . . until we
finally get on to solving the high-level
waste problem in this country. (77)
In summary, the large quantities of low-level
wastes being generated by the ongoing cleanup at
TMI-2 pose substantial challenges in terms of
long-term storage, shipment and disposal. There
are uncertainties regarding the continued availa-
bility of offsite disposal sites. Storage and disposal
of the highly radioactive, damaged reactor core
pose even greater difficulties, given the uncertain
future of high-level waste disposal in this country.
WORKER SAFETY
A principal focus of the Special Investigation
was the safety of workers performing the cleanup
operation. As noted, the immediate threat from
any further accidents at the plant is principally to
the workers. 514 Further, several health physics 3 *
problems have already occurred.
One instance involved violations of health phys-
ics regulations. On March 29, 1979, radioactive
samples were drawn without using standard health
physics protective procedures. This event marked
the first of a series of such problems. The NEC, in
the cover letter to its notice of violation and intent
to impose civil penalties on Met Ed, commented
that "there was a significant departure from nor-
mal health physics procedures and practices.'' (78)
It went on to say, ". . . we believe that insufficient
measures were taken to control health physics ac-
tions and decisions during the course of the
accident." (79)
Then, in late August 1979. five workers perform-
ing maintenance work on a contaminated water
storage system in the TMI-2 auxiliary building
were exposed to levels of beta radiation in excess
of XRC limits. The exposure was to their skin and
extremities. Personnel monitoring the radiation
were apparently not aware of the potential for
high beta radiation. (80)
After a preliminary investigation, the Subcom-
mittee sent a letter to XRC Chairman Hendrie on
September 27, 1979. The letter, referring to com-
ments made by Denton and others, stated :
Mr. Denton also said that "even today we
are still continuing to experience prob-
lems with the utility's [Metropolitan Edi-
son's] attention to health physics." He
expressed specific concern about the util-
ity's "failure to make adequate survey
and [in] understanding the radioactive
environment in which they are operat-
ing." [3S]
Lending further credence to Mr.
Denton's expressions of concern is a recent
XRC critique of the utility's follow-
through to correct radiation protection
problems identified at the site. In an XRC
report obtained by the investigation. D.
R. Xeely. Region I Lead Radiation Spe-
cialist, and J. R. White, Radiation Spe-
cialist, stated that the company "is not
able to effectively administer the radia-
tion protection program commensurate
with the degree of radiological risk which
is presently being encountered. Such risk
is expected to increase as the recovery
efforts expand." (82)
Acting Chairman Kennedy responded to the
Committee, saying :
With regard to a related matter, your let-
ter of September 27 correctly points out
" See pp. 165, this section.
s The discipline of protecting people from unwarranted exposure to nuclear radiation is called health physics.
Health physicists seek to devise means of providing protection. In this instance, these protective steps broke down and
led to safety problems.
"Problems with Met Ed's radiation safety program were highlighted as early as March 20, 1979 (prior to the
accident) when Met Ed contracted with the Nt'S Corporation to perform a study of its health physics program. (81).
175
that the staff has in the past identified a
number of deficiencies in the licensee's ra-
diation protection program which, as yet,
have not been corrected. As discussed in
more detail in enclosure 2, the staff has
been pursuing these matters over the past
several months and will continue to do so.
Neither they nor the Commissioners will
permit expansion of the recovery pro-
gram until these important issues are suit-
ably resolved. Again, the principal con-
cern related to these deficiencies is in
providing adequate protection for the
workers on the site. (83)
In October 1979, GPU Service Corporation be-
gan upgrading its health physics organization and
contracted with Rockwell International to per-
form engineering studies for an onsite health
physics and radiochemical analysis laboratory. It
has been putting together a health physics team
and intended to have the entire organization in
place by the summer of 1980. 37
The NRC also responded to the health physics
problems. Denton organized a "blue ribbon team"
(84) of nationally prominent U.S. health physi-
cists to advise the Commission. Its report, dated
December 1979, outlined the special health physics
requirements of the cleanup and assessed GPU and
Met Ed's radiation protection program. (85) The
team concluded :
The panel confirmed several manage-
ment and technical deficiencies in the pro-
gram. Recent major GPU/Met Ed com-
mittments [sic] and actions demonstrated
a major change in management attitude.
The panel concluded that exposures to
personnel can be maintained to as low as
is reasonably achievable while limited
preparatory recovery work continues and
when further needed improvements are
implemented as needed, the radiation
safety program will be able to support
major recovery activities. (86)
At the same time the NRC panel also stated that
"The present radiation safety program has sub-
stantial deficiencies and requires significant correc-
tive actioH to support major recovery activi-
ties." (87)
Four factors contributed to its conclusions.
First, NRC staff had identified a number of man-
agerial and technical weaknesses in the program.
Second. Robert C. Arnold, a Senior Vice President
of GPU and Met Ed, told the panel that despite all
the comments and recommendations the utility had
received from various sources, including its own
contractors and consultants, it had been unable to
establish an effective radiation safety program.
(88) Third, the panel's interview with utility per-
sonnel at all levels revealed a common feeling that
safety was not respected. (89) Fourth, the utility's
program lacked organization and direction. (90)
The Panel was unable to evaluate the merit of
the utility's plan to upgrade its program. It noted
that "The upgrading of the radiation safety pro-
gram for major recovery activities is not complete.
The Panel cannot judge the capability of this fu-
ture program." (91) The Panel, however, did state
that the management of GPU and Met Ed had
demonstrated a strong commitment to improve the
radiation safety program and that work could pro-
ceed safely on a limited basis under existing
management. (92)
The significant problem of protecting workers
will continue. The number of technicians at the
site is far greater than during normal operations.
Usually, plant personnel, even during a refueling
outage, number only about 800. 38 As of April 1980,
there were about 1,100 people onsite to work on
TMI-2. Sixteen contractors were involved in onsite
radiation protection alone, and more than twice
this many contractors were onsite for all purposes.
(93) There were also numerous representatives of
various government and industry groups.
In addition, the magnitude of contamination and
radiation with which individuals must work is
many times greater than previously faced. Further,
there is an unusual aspect to the situation at TMI
a preponderance of beta radiation, a result of the
cesium-137 and strontium-90. It is more common,
around operating reactors, to have predominantly
gamma radiation.
A team of three health physicists was scheduled
to enter the containment in mid-1980. They would
be exposed to this beta radiation. If the krypton in
the containment were not vented, the field would
be expected to be approximately 400 rad/hr. If it
were vented, it would be significantly less. The
actual dose received would also depend on where
in the containment the team goes. Exposure could,
in any case, be limited by wearing layers of special
protective clothing, and only the extremities would
receive significant doses.
Protective clothing does not block penetration
by most of the gamma radiation at the Unit 2
facility. The level of this radiation, however, was
estimated to be low enough that workers could
tolerate it for a few hours at a time. Thus manned
entry into the highly radioactive containment was
37 As of December 1970, the organization consisted of 194 people : 55 assigned to dosimetry, 89 to field operations,
15 to radiation technical support and 35 to bioassay and management.
38 Refueling is done periodically to replace fuel that has reached its design life and to reshuffle good fuel in the
core. During refueling of the reactor, many more people than normal are onsite. Normally, TMI-2 has around 200
people onsite.
176
"Trailer City" : Temporary offices set up during the accident have been needed for recovery operations
considered possible if sufficient care were taken
and the period spent there was brief.
When asked during the November 8 hearing
what the total radiation dose to workers would be
for the entire recovery, Wilson of GPU Service
Corporation said :
We don't yet have a total estimate of what
we might expect of what I would charac-
terize as a total man-rem dose of the re-
covery operation. I would expect we
would anticipate there would be no work-
ers in the absence of any incident that
would receive what's called a maximum
allowable dose. (94)
THE TECHNICAL STEPS
MAINTAINING STABILITY
Subcriticality must be maintained in the core to
prevent possible release of fission products or melt-
ing of the core. Subcriticality is maintained by
keeping the core covered with water that has an
adequate concentration of boron. As long as the
primary cooling system remains intact, it is un-
likely that the water level will decrease to the
point of uncovering the core. Further, some con-
trol rods may not have melted ; if they are in place,
they will help insure shutdown.
Maintaining Boron
Boron in solution can surround a core whose
normal geometry has been lost. Calculations show
that between 3,000 and 4.500 parts per million
(ppm) of boron in the water will insure Subcrit-
icality no matter what the geometry of the
core. (95)
As of March 1980, a concentration of around
3,500 ppm was being maintained, controlled
through the use of the Borated Water Storage
Tank and the make-up and let -down systems. The
sample of the water from the primary system is
analvzed weekly for boron, gross radioactivity and
acidity. GPU has stated that as long as the reactor
fuel and control materials remain in their present
abnormal and uncertain arrangement, the boron
concentration must be maintained within that pre-
scribed range to guarantee Subcriticality of the
core and to avoid excessive heat. (96)
177
Cooling the Core
Decay heat removal is achieved by circulating
coolant around the core. This heat naturally de-
creases with time. At the end of April 1979, it was
41 million watts ; by April 1980, it had dropped to
180,000 watts, 39 still enough to heat around 7 homes
on a winter day. If this heat were not removed, the
reactor would heat up a few degrees Fahrenheit
per day until the balance point referred to earlier
was reached. 40
As of April 1980, natural circulation was the
method being used to remove the heat. The water
was circulating only within the containment; as
the building was still sealed, the chance of spread-
ing contamination or releasing radiation was re-
duced. Further, natural circulation lessened the
exposure of workers to radioactivity.
On the primary side, natural circulation in-
volved onlv the core and steam generator "A" ; no
pumps were being used. On the secondary side, the
principal pieces of equipment relied upon were
the condenser and f eedwater system. Together they
turned the steam produced by the heat absorbed
from the primary system into water and pumped
the f eedwater back to the steam generator, where
it again absorbed heat from the primary side. For
"steaming," as this process is referred to, to take
place, the primary system, including the piping,
seals and pressure vessel containing the core, must
remain intact.
If difficulties should arise with natural circula-
tion, two back-un cooling systems are available
immediately. A third svstem was scheduled to be
ready for operation in March 1980. 41 (97)
According to GPU, one such difficulty would be
a leak in steam generator "A." (98) If it was be-
tween the primary and secondary sides of the ven-
erator, radioactive contamination could get into
the water flowing through the secondary svstem,
creating additional radiation hazards and increas-
ing the potential for contamination spreading
through leaks in the secondarv svstem. Because of
these hazards, other means of cooling the reactor
would be sought and the steam generator shut
down.
Steady cooling through natural circulation could
be. disrupted by temperature changes. As of April
1980. the bottoms of the steam generators were
submerged in the pool of water in the containment.
As rlecav heat decreases with time, the coldest point
in the svstem will shift location, leading to fluctua-
tions in the flow paths of the coolant. These oscilla-
tions, which would take place in the primary loops,
could lead to lesser oscillations of flow in the core,
with a resulting fluctuation in core temperature
of a few degrees Fahrenheit. Although the
amount of this temperature fluctuation is very
slight and tends to be self -correcting, it is un-
desirable because prudence dictates that steady
cooling should be maintained. 42
When coolant flow becomes unstable, temporary
stagnation of the flow results, potentially leading
to boron stratification differing concentrations of
boron in the water at various levels or regions of
the core. According to GPU, calculations indicate
that the fluctuation in boron concentration from
this phenomenon is onlv minor, and concern
over any resultant recriticality is unwarranted.
(100)
The NRC and GPU indicated that boron strati-
fication was not expected to be a problem. (101)
since the utility could provide adequate flow to
mix the concentrated boron solution injected into
the reactor coolant system so that the stagnant
regions would not become depleted. (102)
Backup Cooling Systems
The first back-up system would involve the "B"
steam generator. It uses water flow as opposed to
steaming water is pumped through the svstem
and absorbs heat from the primarv side with con-
ventional heat exchangers. The leak that was iden-
tified in the "B" steam generator during the acci-
dent apparently has not recurred, and the "B"
generator could be used. Because this option takes
place in part outside the containment, it is a less
desirable method for cooling, as it increases the
possibilitv of spreading contamination in the plant
fnd exposing workers to higher levels of radiation.
The risk of a release of significant radiation to the
environment is minimal with this system.
The second back-up method involves the regular
Decay Heat Removal System, which also operates
in part outside the containment. It involves pumps
and heat removal equipment and is the one nor-
mally used after a reactor shutdown. It, too, is not
being used so as to limit the spread of contamina-
tion and exposure to workers.
The third alternative the Mini Decay Heat
Removal Svstem had been scheduled for opera-
tion in March 1980. but was still not being used as
of June 1980. In that month the NRC nnd GPU
decided to install a long-life filter. (103)
39 Heat and electricity are both forms of energy. The rate of energy production can be expressed in watts.
" See p. 166.
11 This system the Mini Decay Heat Removal System- will eventually become the principal means of cooling ; it
is discussed in detail later. In order to brine this system into operation, GPTT Service Corporation has had to submit
to the NRC proposed changes in the Technical Specifications of its license. The NRC must approve these.
42 In April 1980. such oscillations in flow actuallv occurred. Thev resulted from the occasional use of the so-called
"pressure-volume control system." The system caused just enough imbalance to start the oscillation. The oscillations
occurred over a period of 17-18 hours and led to a maximum temlperature drop of 40-50F across the core during
periods of complete stagnation. Eventually, the oscillations disappeared and steady cooling was re-established. (99)
The importance of these oscillations has not been thoroughly investigated.
178
This system is smaller than the regular decay
heat removal one, as it has been built to accommo-
date the low levels of decay heat being generated
in 1980. If nothing unusual happens, it will be
more than adequate to handle existing levels of
heat.
The Mini Decay Heat Kemoval System also
functions in part outside the containment. How-
ever, it has much smaller piping and other equip-
ment than the regular Decay Heat Removal Sys-
tem, and the volume of radioactive material to be
pumped through it will be smaller. The associated
radiation hazard would decrease proportionately.
Nearby workers would be exposed to significantly
less gamma doses than result from the regular
system, which would involve several hundred rem/
hr. Further, use of the Mini System would allow
external control and monitoring of coolant condi-
tions (both temperature and pressure) and permit
access from outside the containment to the primary
coolant, a capability that will be important during
cleanup of primary system water.
Construction of the Mini Decay Heat Removal
System is the first of a series of actions that would
establish independent, external control of the re-
actor environment. According to Wilson, it "will
be the mode of cooling until such time as the [re-
actor] head is removed and the fuel extracted."
(104) GPU plans to rely on this system because,
as noted earlier, the utilitv believes that natural
circulation could be impeded in the future. 43 More-
over, once the reactor head is removed, natural
circulation will no longer be possible, and a dif-
ferent method of cooling must be used.
EARLY CLEANUP STEPS
Major decontamination of the plant began in
April 1979. when Met Ed started processing the
water that had been transferred from TMT-2 to the
TMI-1 auxiliary building for storage. The proc-
essing was done with a system called EPICOR-I. 44
The water contained relatively low levels of radio-
activity (less than one microcurie per milliliter).
In April, decontamination of the diesel genera-
tor building was undertaken. In May, work was
begun at the TMI-2 auxiliary and fuel handling
buildings. It involved nrincipallv dry and wet vac-
uuming, mopping and wiping to remove contam-
ination. "Workers were required to use special
clothing and respirators.
The accident and cleanup as of April 1980 al-
ready had produced a variety of slightly radio-
active solid wastes, such as clothing, rags, ion-
exchange resins, swipes and contaminated air
filters, much of which has been buried at the Rich-
land. Washington site. (105)
Radioactive Water in the Auxiliary Building
Cleanup of this building has been progressing
well. (106) As a result of decontamination efforts,
surface contamination in selected areas had been
reduced by a factor of between 100 and 1,000 be-
tween May and October 1979. Since October, more
of the building has been cleaned in order to achieve
levels of radioactivity comparable to the already
decontaminated areas. Although the water in one
tank in the building was still reading more than
1,000 rad/hr, most of the others were reading from
20-30 rad/hr. (107) Processing of this water is
being done with EPICOR-II. 45
The cumulative exposure of workers in the aux-
iliary building through November 1979 was 50
person-rem. (108) By comparison, approximately
45 person-rem are received per year by workers
at a plant which is being refueled. (109) The
amount of radioactive water in the auxiliary build-
ing had been increasing prior to November 1979 *
because of non-radioactive water leaking into the
system. While the additional non-radioactive water
lias had a diluting effect, it too has become con-
taminated, thereby adding to the volume of water
needing to be processed.
The leakage is mostly from pumps in the build-
ing that are part of the river water service sys-
tem. (110) At present, these pumps cannot be
sealed because they are too close to high-radiation
areas. Prior to November 1979, the amount of
leakage had been fluctuating from as low as 300
gallons per day to as high as 2.000 gallons (the
average was 800-1,000 gallons/day), depending on
the use of various systems on any dav. By April
1980, the leakage was better controlled, and the
average ranged between 200-450 gallons per day.
(111) The water drains into the contaminated
sumps in the auxiliary building where it is col-
lected and added to the new tanks that were in-
stalled in the fuel handling building to contain
contaminated water.
The total usable capacity of the storage tanks
to contain this leaking water is 415.000 gallons.
(112) One week after the accident, the design of
the EPICOR-II water processing system was be-
gun, in part to address the growing problem of
storage of radioactive water. The system was
scheduled for use in mid-May 1979. At that time.
th city of Lancaster. Pennsylvania, brought
suit 4T to gain an injunction against the discharge
See p. 178.
" Similar in design to EPICOR-II, EPICOR-I was brought to the site immediately after the accident.
See pp. 181. 182.
* In November 1979. with EPICOR-II operating, the water could be processed faster than it was leaking in. and
the total amount of radiation was being reduced.
<T See "Legal and Regulatory Aspects of Recovery." pp. 201-204.
179
Decontamination underway
of processed water into the Susquehanna Eiver.
(113) (Lancaster is about 23 miles from TMI and
gets its water at a point 8 miles downstream
of the site.) Another suit was filed by the Susque-
hanna Valley Alliance against both the utility and
the NRC. (114) As a result, the NRC prohibited
any further water processing and discharge with-
out its authorization. The NRC's action raised a
potential problem of storage of the contaminated
water.
The Subcommittee recognized the need for a de-
180
cision on the issue of water storage. On Septem-
ber 28, 1979, it sent a letter to the Chairman of the
NRC, stating:
We understand that the currently esti-
mated capacity for storing this water in
Unit 2 will be exceeded in approximately
40 days. We understand that alternate
storage options exist, including pumping
the contaminated water into tanks in Unit
1 or bringing additional storage tanks on
to the Island. Please advise us what op-
tions are being considered and how they
would be implemented.
* * *
We bring these matters to your attention
because of the serious public policy issues
they pose, not only for the Three Mile
Island region, but also with respect to
XRC's ability to handle this matter. (115)
By October 1, only 28,600 gallons of storage
capacity remained iii the tanks installed in the
fuel handling building. On October 16, GPU said
that approximately 23,000 gallons of storage ca-
pacity remained. (116)
The XRC. after receiving the Subcommittee's
letter, held a meeting on October 4, 1979. XRC staff
told the Commissioners that any significant fur-
ther delay in decisionmaking could lead to several
problems'. (117) First, the tanks could overflow,
spilling water into the auxiliary building sump,
from which it would flow back into the full sump
tank system and ultimately begin to fill the build-
ing. Second, the contaminated water from Unit 2
might have to be transferred to the uncontami-
nated tanks in the TMI-1 unit (the two units share
certain water storage facilities), thus spreading
contamination to Unit 1.
Possible solutions besides further processing in-
cluded acquisition of additional tanks or pumping
the water in the auxiliary building back into the
containment (118) Both options had serious draw-
backs. (119)
On October 22, the NRC decided to permit Met
Ed to process the contaminated water with
EPICOR-II. 48 However, the Commission also de-
cided that the processed water would have to be
held up pending a later decision on disposal. (120)
The clean tanks, also located in the nearby fuel
handling building, now hold processed water.
Because the EPICOR-II system was basically a
new design, the extent to which it could remove
radioactive contamination was uncertain. 49
" See "Legal and Regulatory Aspects of Recovery." pp. 201-203, 206-207. concerning the claim that the XRC illegally
"segmented" cleanup decisions in violation of the National Environmental Policy Act of 1968 i and other regulations.
Thi* capability is expressed by a quantitative measure of effectiveness called the decontamination factor for
the process ; that measure is the ratio of the concentration of radioactivity in the contaminated water to that in the
processed water.
EPICOR-II water purification syste
181
In testimony on November 8, Wilson told the
Subcommittee that the decontamination factor
for EPICOR-II was better than anticipated :
The decontamination factors [of cesium
137] are about two orders of magni-
tude [50] better than the design basis of
the system. We expect to be able to con-
tinue to process with this system and are
putting in place on the site the additional
storage tankage for the clean water. (121)
He also said,
We are not now, or do we have immedi-
ate plans to discharge that water. In fact,
they are under a probation [sic] from the
NRC to not do so. (122)
THE NEXT STEPS
The Containment Atmosphere
The next step in recovery of Unit 2 is removal
of the estimated 45,000 curies of krvpton 85 in the
containment atmosphere. (123) GPU Service Cor-
poration has considered several options, including
controlled venting to the atmosphere, cryogenic
processing, charcoal adsorption, and gas compres-
sion. In addition, selective absorption has been
proposed. Venting has become a very controversial
step, as it involves releasing the gas to the atmos-
phere. While the last four options would avoid
direct releases, they have other drawbacks, as
described below.
Controlled Venting
In the first option, the gas would be released
from the building by venting it at a controlled rate
over 34 days, through the plant vent stack, 160
feet above ground level. Venting would take place
at times when wind and other meteorological con-
ditions are most favorable for atmospheric dis-
persion. (124)
GPU has maintained that this controlled release
can be performed in compliance with all current
Federal radiation standards. 51 (125) Its estimate
is that the highest calculated dose to an individual
would be 0.1 millirem of gamma and 5 millirem of
beta radiation for the total purge (the standards
are 10 and 20 millirem, respectively) , 52
The NRC discussed the magnitude of the release
at a meeting on November 29, 1979. One issue was
comparative doses : how much radiation would be
released in the venting at TMI-2 in comparison
with normal releases from either a pressurized
water reactor (such as TMI) or a boiling water
reactor. 53
The NRC staff told the Commission during the
meeting that the controlled venting of krypton
from the TMI-2 plant would have fewer radiologi-
cal consequences than do the releases of krypton
and all other noble gases M over 1 year from a
single, normally operating boiling water reactor.
(126) The release would also be 10 times less than
the annual routine releases from certain Federal
military installations. 55
On August 22, the Critical Mass Energy Proj-
ect, a public interest group that opposes nuclear
power, petitioned the NRC to prevent the con-
trolled venting. Richard Pollock, its Director, sent
a letter, dated August 22, 1979, to Chairman Hen-
drie, saying:
According to the Bechtel Corporation
consultant's report for the licensee, "con-
trolled" venting of radioactive gases could
lead to contamination levels for persons at
the boundary site reaching .14 millirems
of gamma radiation and 14.8 milli-
rems [56] of beta radiation during a 30-day
period. NRC criteria sets the yearly maxi-
mum dosages for the general population
at 10 millirem for gamma radiation and
20 millirem for beta counts.
* * *
... if there was an accident during vent-
ing, the TMI-2 area residents conceivably
could receive much larger dosages than
those contemplated by Bechtel and GPU.
(129)
60 Two orders of magnitude equal a factor of 100.
" 10 C.F.R. 60, Appendix I, B.I, states "The calculated annual total quantity of all radioactive material
above background to be released from each light-water-cooled nuclear power reactor to the atmosphere will not result in
an estimated annual air dose from gaseous effluents at any location near ground level which could be occupied by in-
dividuals in unrestricted areas in excess of ten millirads for gamma radiation or 20 millirads for beta radiation."
"The safety analysis discussed here pertains to an estimated inventory of 44,000 curies. In its environmental
assessment the NRC used a figure of 57.000 curies, based on weekly sampling of the reactor building atmosphere since
the accident. The licensee's figure (44.000 curies) is based on the measured concentration at the time its report was
issued (November 13, 1979). The conclusions reached by the licensee would probably not be significantly different if
the higher figure were being used.
" See "Technical Glossary," Appendix E, pp. 367, 373.
54 Noble gases routinely escape from reactors and include helium, neon, argon, krypton, xenon and radon.
* For example, the Savannah River facility located in South Carolina releases 4.3 X 10" Ci of Kr-8;5 per year. The
maximum calculated whole body dose to an individual at the plant perimeter from the krypton is 0.0026 millirem. (127)
Standards for those facilities are set by DOE.
"These radiation dose predictions were based on preliminary estimates made by Bechtel Corp. in its July 1979
Planning Study for containment entry and decontamination. (128) Hence, they should not be expected to agree with
the later estimates cited above
182
illation doee from
The nragedni per
dental relentK is uncertain, as it depends on the
extent of dte leak and prevailing wind and other
tArr conditions. However, according to Voll-
r who spoke at a meeting of tin XRC on Xo-
29,1979:
... if TOU are involved in an accident
where yon released all die krypton cur-
rently in containment, using average
wteorougy, you would get a whole-body
done in tie environment of less dtan 10
MR [mfflirem] and a skin dose on the or-
der of 200-500 MR. so even if all [the
krypton] were released in an accident
case, die offsite consequences would be not
large even with respect to perhaps Part 20
[die Federal Guidelines]. (110)
Hence, although diere is strong public distrust
about venting, XRC estimates indicate minimal
efects for the surrounding population.
even in die event of an accidental release at one
time of all the krypton in the containment. (131)
Because the discussion of venting has led to con-
siderable public pmbuie to evaluate other options
< see -Social Issues in Recovery " pp. 199-200. 201 ).
Governor Tbornburgh asked that the Union of
Concerned Scientists pet faint an independent
studv. It concluded. " direct radiation-induced
health ejects from exposure to krvutou 85 even
from the Met Ed >TRC proposed venting would be
absent." (132) However, die report stated, ~TCS
nrmftfnA* mgmm** any procedure that would re-
sult in iltiaum . . . being deliberately exposed to
at levels comparable to those ex-
. . dw venting proposal. (1-
e to <KmiiiiA public stress, FC>
osal? for elevated venting:, each of
pq*rf*l of greatly reducing die
sure. One option involved construc-
inerator stack to create a buoyant
t other involved use of a tethered
t an vfjtimAftl stack made of thin
pected f
Citing a
made tw
which h
radiatio
tion of
::/:. -
DO! veth vlenel one foot in diameter. UOOO-&000 ft.
high. (13*)
Cryogenicl
The second option for krypton removal involves
cryogenic tmeessine. The *Kr-R5 would be lique-
fied, distilled and stored in bottles. An advantage
of this method is that it could significantly de-
crease or eliminate the radiation released from the
plant at the time of removal. However, the utility
has estimated that it would take 20 to 30 months
to buQd the necessary equip- ?>5) Further.
it is unclear what would be done with the bottles.
A safety analysis and environmental report pre-
pared by GPF outlined the disadvantages :
IV
s highly concentrated
[Kr-85l. Any leakage or component fail-
ure could result in significantly greater
amounts of uncontrolled radioactivity
release than die other systems. (136)
* * *
There is no significant operating experi-
ence with a cryogenic distillation system
at any operating light water reactor. Ac-
dinglv, this is not a proven technology
:of
for reactor application. (137)
The study concluded dot
When compared to controlled piugwg u*
the containment building, the alternate
cryogenic treatment system is considered
to* be less safe it is less reliable, and
clearly has the potential for uncontrolled
releases of radioactivity with higher radi-
ation exposures. (138)
During the November 29 XRC meeting, die
issue of cryogenic cleanup was also addressed.
XRC estimated that if this option were selected, it
would take 20 months to become operational after
a decision was made to proceed. (139)
Charcoal Adsorption
The third option, charcoal adsorption, could also
radiation releases from the plant, Dur-
ing the same Commission meeting, Vollmer ex-
plained its use :
The technology . . . is one simply of put-
ting the contaminated gas over charcoal,
preferably in a chilled state, preferably
under some bigher-than-atmosphere pres-
sure, to get maximum effectiveness, and
then the charcoal which would adsorb the
krypton gas, but not retain it indefinitely,
would have to be encapsulated and then
you would have to [dispose of it]. (140)
Yollmer continued:
I might indicate that the charcoal volume
required for this would be about the same
as the size of the volume of the contain-
ment building, about two million cubic
feet. What the staff tells me is something
on the order of a third of the charcoal
available in the country. (141)
According to GPF. the fourth option, gas com-
pression, similarly would greatly lessen die
amount of krvpton-85 released into the atmos-
phere. On die other hand, as the GPF safety anal-
ysis report noted :
Storage of krypton at high pressure for
long periods of time in 28 miles of piping
the likelihood of uncontrolled
npared to purging containment.
The extensive time required to build and
install a gas compression system would
increase the likelihood of inadvertent and
uncontrolled leakage from the existing
183
containment building, and thereby cause
higher exposure to personnel. (142)
The report concluded :
When compared to controlled purging of
the containment building, the alternate
gas compression system is considered to
be less safe it is less reliable and clearly
has the potential for uncontrolled release
of radioactivity with higher radiation
exposures. (143)
This option would require construction of a $50-
$75 million facility over a two- to three-year pe-
riod. (144) The facility would include a 160-foot
high building to house the equipment and approx-
imately 24 miles of 36-inch diameter high pressure
piping. (145)
Selective Absorption
In a briefing before the NRC on April 25, 1980,
representatives of the Department of Energy and
Oak Ridge National Laboratory presented a tech-
nical assessment of an alternate method for treat-
ing the krypton. (146) The selective absorption
method, instigated four years ago by a DOE re-
quest that a mobile radwaste disposal system be
developed, is partly a spinoff from technology de-
veloped for the removal of krypton from reproc-
essing facilities.
A report written by Oak Ridge National Labo-
ratory (ORNL) estimates that development of
such a system for use at TMT-2 would require 13
months and from $9 to $12 million, on a crash pri-
ority basis. (147)
In testimony before the Commission, ORNL
stated that it believed venting was the preferred
option. (148) The selective absorption method is,
in principle, a zero-release svstem according to
ORNL. but venting would still be preferable since
it would allow early entry into the containment.
(149)
The selective absorption method was independ-
ently assessed in two other studies. A professor at
the Michigan State Universitv wrote in a letter
to Commissioner Gilinsky. "I have tentatively
concluded that the best method of those available
is the selective absorption process system. . . ."
(150)
Science Applications, Incorporated, also per-
formed a review and found that there was little,
basis to choose between selective absorption and
controlled purging with regard to phvsical health
effects and that the purging option was preferred
because it would be less stressful to the population
than the selective absorption method. (151)
.
NRC Consideration of Options
On March 12, 1980, as planned, the NRC staff
submitted their environmental assessment of the
venting option. They concluded that controlled
purging was the preferred alternative. They pro-
posed a period for public review, to be followed by
a meeting at which the public's views would be
heard. Thereafter, the Commission was to make a
final decision on venting. 57
Containment Cleanup and Core Removal
The containment must be decontaminated so that
it can be entered and the highly radioactive core
dismantled and removed. The water in the contain-
ment also poses a health physics and safety
problem.
Containment Entry and Water Removal
The plans outlined in the July 1979 Bechtel re-
port considered the use of robots for entering the
containment. (152) The utility has since concluded
that radiation levels will be low enough to allow
manned entry for brief periods. (153)
A team of three health physicists, two GPU
employees and three backup members, was trained
for this job. The individuals had been selected on
the basis of their knowledge of the layout of the
containment and of health physics, understanding
of the operations to be performed and physical
fitness. The team will carry out a number of tasks,
such as surveying for radiation, assessment of con-
tamination and observation of the conditions in-
side the containment. Their analysis will be the
basis for planning further entry and decontamina-
tion. The team had completed its training by mid-
March 1980 (154) and was awaiting NRC ap-
proval of its procedures.
On May 20, 1980, workers encountered an
unexpected problem when they attempted to enter
the containment for a 30-minute inspection. Entry
was thwarted after 15 minutes of effort when the
containment door beyond the eouipment airlock
failed to open. (155) The door depends on func-
tioning of an electro-mechanical system. As of
June 1980, the NRC had not determined whether
the door was stuck because of the failure of elec-
trical equipment, mechanical equipment, rusting
inside containment, or combinations of these. Test-
ins: on a similar airlock door was underway in
order to attempt to identifv the problem.
After the team conducts its assessment, the next
step will be to clean up the radioactive water
within the containment and the reactor. 88 The util-
ity is planning to process these highly radioactive
liquids (between 100 and 275 microcuries per cubic
centimeter) using a submerged demineralizer that
17 See "Social Issues in Recovery." pp. 199-200. 201, and "Legal and Regulatory Aspects of Recovery," pp. 206-207, for
a discussion of events subsequent to the staff's venting recommendation.
58 Although the core will continue to leak radioactive contamination into the coolant, cleaning the water will greatly
reduce the amount of radiation and thus lessen exposure of workers to it.
184
was originally developed for defense application.
(156) The system is much like EPICOR-II, but
uses inorganic rather than organic resins. It will
be placed in the spent fuel pool where the water
will shield it from the radiation.
In testimony before the Subcommittee on No-
vember 8, Wilson said that specialized engineering
and development of this system for use at TMI is
underway. He said it is expected to be operational
in late 1980. (157) One existing system that might
be suitable for use at TMI consists of a self-
contained unit within a shipping cask that is li-
censed for shipments of up to 300,000 curies. It
would allow direct shipment of the wastes, thereby
minimizing handling. (158)
An evaporation system will be required to clean
up the liquids used in decontamination. These
liquids contain a soap-like material on which
EPICOR's ion exchange technique does not work
well. The evaporation system being procured by
GPU Service Corporation can process 30 gallons
per minute. It boils the water; the steam is then
condensed and sent through a polishing demineral-
izer 59 designed to produce very pure water. This
water will be reused during decontamination.
As noted, there were 700,000 gallons of radio-
active water in the containment as of December
1979, a level that has been increasing because of
the leaking pump seals inside the containment. In
testimonv before the Subcommittee on Novem-
ber 8. Vollmer said :
. . . [given! the leakage rate of about 500
gallons per dav, and I believe it's actually
lower than that now, we would -project ap-
proximately a foot or so rise in six
months. (159)
A more precise estimate by GPU established the
rate of leakage at less than 230 gallons per day,
equivalent to a one-half to one inch increase in the
water level per month. (160) The leaking cannot be
reduced until the new primary coolant pressure
control co and heat removal systems are operable.
The rising water, as noted, threatens the motors
that operate two criticnl isolation valves. Bflsed
on photographs taken before the accident that
show the height of the motors, it was calculated
that, as of November 1979. the motors were ap-
proximately two feet above the water level. In
November Vollmer told the Subcommittee :
... I don't see anything in the near term,
say within a year, that would have any in-
fluence on the safety of operations. (161)
By April 1980, the bottom of the valve bodies
were in contact with the water 61 and a failure of
the valve, given the humid environment, is possi-
ble. If the valves become submerged or if the elec-
tric actuators fail in the humid environment, (162)
control over them will be lost, and they will remain
in whatever position they were prior to failure. As
they have been kept closed, they would fail in that
position. Access to primary system coolant would
then be lost, and neither the decay heat removal
system nor the new Mini Decay Heat Removal Sys-
tem could be used. This would be undesirable be-
cause it would leave only the existing deteriorating
equipment for core cooling. In addition, decon-
tamination of the primary system water would be
greatly inhibited.
Met Ed frequently measures the water level and
continuously monitors (meggers 62 ) the electrical
leads on top of the valves. (163)
Since it is important to the long-term cooling of
the core and treatment of the primary system
water to have one of the valves open, if difficulties
arise, the utility will open one of the two valves so
that it will fail in that position. If such action has
to be taken before the NRC approves the Mini De-
cay Heat Removal System, 63 radioactive water will
flow into the regular decay heat removal system,
spreading contamination in the facility. That
spread can be limited by closing an isolation valve
located outside the containment building.
There also is radioactive material within the
reactor vessel and its associated piping. The
amount continues to increase as the coolant moves
around the damaged core and fission products in
the fuel are leached from the fuel. (164) This proc-
ess is counteracted to a certain extent by continual
replacement of the leaking radioactive water with
clean water, which dilutes the radioactivity some-
what. The activity level in the primary system
water is now 275 microcuries per milliliter, essen-
tially the same as that in the water in the contain-
ment.
"A polishing demineralizer is the final filtering stage of the evaporator system. The relatively decontaminated
water is "polished" in this final step.
* The coolant pressure control system located in the fuel handling building is also used to provide make-up coolant
to compensate for the leakage and to control the chemistry of the coolant, particularly the concentrations of boron and
oxygen. It includes a pressurized nitrogen supply that controls pressure, and a borated water batching tank, charging
water storage tank and independent charging pumps. It was added for the same reason that the new heat removal sys-
tem WRS added : to avoid reliance on reactor systems that may have been damaged during the accident and that are in-
acceihle 1-erause of their proximity to high radiation fields.
n The valve bodies are about one foot below the actuators.
K Meggering is the electrical monitoring of the leads. Presence of water will lead to a change in the electrical
signal, which would be detected on a readout instrument, alerting personnel to possible degradation of the equipment
electronics.
* According to GPU. if the XRC requires a long-life filter on the Mini Decay Heat Removal System, four to six
months will be required for installation.
185
5U-058 0-80-13
Because of the continued leaching, GPU Service
Corporation intends to install continuously operat-
ing purification equipment, once the reactor head
is removed, to minimize the dose to workers. (165)
Building Decontamination
The next step in the cleanup process is manual
decontamination of the inside of the containment.
A containment recovery service building will be
constructed adjacent to the containment equipment
hatch to provide health physics control and isola-
tion from the environment.
Many techniques are likely to be employed in
this stage of recovery, including wet and dry
vacuuming, mopping, wiping and the use of semi-
portable equipment such as degreasing units, ultra-
sonic cleaners and electropolishing machines.
(166) Conventional contamination control tech-
niques are the technological basis for this activity.
A decision on specific methods will be made once
the observations and measurements taken in the
earlier manned entry are analyzed.
Radiation Inside the Containment
The work force will face the greatest radiation
hazard in this phase. Early after the accident, the
NEC and the utility had been concerned that
cesium 137 (Cs-137) might have been released
from the water into the atmosphere of the contain-
ment and then might have become extensively de-
posited on the walls of the containment, 64 adding
significantly to the cleanup task. (168) The radia-
tion dose to workers would have made the health
physics problems substantially more difficult, and
it would have been necessary to wash the Cs-137
off the walls remotely, using the building spray
system in conjunction with solvents and deter-
gents or foaming agents.
Information obtained prior to the Subcommittee
hearing on November 8 showed that the extent of
airborne Cs-137 and of surface deposits was much
less than anticipated, since less cesium was released
into the air than estimated earlier. This means that
the cesium is in solution in the water and can be
cleaned up by other techniques such as EPICOE
or the submerged demineralizer.
As of June 1980, GPU's plans called for manual
decontamination of the walls. In remarks before
the Subcommittee on November 8, Wilson said,
... a part of the plan originally con-
ceived by Bechtel used the containment
building spray system as a means of re-
mote decontamination inside containment
prior to entry. The current data suggest
that's not required. . . . (169)
The feasibility of this plan will depend on sur-
veys performed by the team in the containment.
Although attempts at direct measurements have
been inconclusive to date, (170) preliminary in-
dications of the condition of the containment wall
have been obtained. A two-inch and a nine-inch
hole were drilled through the end plates of the ac-
cess pipes leading into the building. (171) As of
June 1980, Oak Ridge National Laboratory had
finished analyzing the cut-offs from the holes. Re-
sults have not yet been released. Swipe tests 65
around the inside perimeter of the hole and radia-
tion surveying are also planned. These tests will
help determine the concentration of Cs-137 near
the hole, while the analysis of the cut-outs will be
used in determining the efficacy of various decon-
tamination techniques.
Finally, the inside of parts of the containment
has been videotaped, using equipment inserted
through the holes. According to the NRC, the pic-
tures show no significant damage or disruption as
a result of the hydrogen burn that occurred on
March 28, 1979. (172) No visible evidence of the
accident can be seen on the portion of the contain-
ment filmed except for some paint blistering and
droplets of condensed water falling from the walls.
On the basis of the limited measurements avail-
able, the utility has estimated the radiation
environment inside the containment building;
(173) its figures are shown in Table 2. The figures
in this table were estimated based on the assump-
tion that neither the radioactive gases nor the
liquid inside the building had been removed. 66
TABLE 2. Estimated dose rates at various elevations
in reactor building
Dose rate (rad per hour)
Elevation Location
Total
panuna
Total
beta
Total
282ft Sump 120.0 720 840
305ft Equipment hatch.. 3.0 400 400
347ft Operating deck .5 400 400
If the krypton and water are not removed first,
manned entry will be possible only for relatively
short times. Because of the estimated high level of
radiation, containment cleanup (krypton and
water removed and walls decontaminated) must
be finished before the reactor head and core can
M This is similar to what happens when soot adheres to buildings, discoloring them. While there is no evidence
that cesium has migrated into the steel liner of the building, there is still a concern that some painted surfaces may
have to be stripped to remove cesium contamination. (167)
M A swipe test is a means of determining the level of contamination on a radioactive surface. A small piece of
cloth-like material is rubbed on the wall and taken to the laboratory for radiochemical analysis. The test allows deter-
mination of the relative amounts of radioactive species present.
66 Values in the table were current through March 1980.
186
be removed. The estimated radiation dose will come
predominantly from gamma and beta radiation
and will differ according to the elevation within
the containment The radiation environment at the
282-foot level is primarily the product of radiation
emanating from the pool of water. At the higher
elevations ( 305 and 347 feet ) . it is largel v the prod-
uct of emissions from the krypton. If the contain-
ment sump is drained and the krypton 85 removed,
gamma dose rates inside the building are projected
to drop sharply to between 0.2 and 10 rem/hr.
For purposes of comparison, the occupational
whole-body dose limit for an individual, as estab-
lished by the Code of Federal Regulations, is 3 rads
per quarter year from all sources. (174) The dose
permitted to* the hands and feet is approximately
18 rads per quarter year (hands and feet have a
greater tolerance for radiation). (175)
Using the radiation estimates, the amount of
time that an individual can remain in each area of
the plant without the aid of additional shielding
can be determined. If. for example, the gamma
radiation level were 3 rem/hr, an individual could
work no longer than 1 hour in that region.
Reactor Head and Core Removal
This phase of the cleanup is the most uncertain,
since the condition of the severely damaged core
is unknown and because subcriticality of the re-
actor must be monitored and maintained while
work proceeds on the core. A plan for removing
the reactor head " and core was being prepared
by Bechtel. (176)
Barring legal and economic difficulties. CPU's
goal was to begin removing the head 11 months
after the containment was entered and to begin
removing the fuel 20 months after entry. The
XRC's preliminary estimate, based on CPU's
plans, was that core removal might not be complete
until March 1984 if the krypton in the containment
were treated crvogenically instead of being
vented.
The plan for removal of the head and core has
been slow to develop. Certain steps are funda-
mental to the job. but the specific techniques to be
employed will depend upon now uncertain details,
(177) For example, it is clear that the reactor
head must be removed to gain access to the core.
It is less clear, however, whether special tools and
procedures will have to be developed to disengage
entangled control rod drives. That will only be-
come clear once the containment can be entered
and tests performed. The special tools can only be
designed once the nature of the problem is
defined.
According to GPU. the Bechtel report on the re-
moval of the head (which was released in May
1980) contains initial planning for the removal of
the reactor vessel head and core, describes some of
the available technical options and also identifies
the preferred general approach to the job.
Two steps are necessary before the reactor head
can be lifted off. First, a means of removing decay
heat other than by natural circulation must be
established, since natural circulation will not work
with the reactor head removed. As noted, the Mini
Decay Heat Removal System will probably be
used. Second, the coolant must be decontaminated
and a continuous filtration system hooked up in
order to reduce the radiation field for workers.
Once the containment building and primary water
are decontaminated, access to the reactor head
area needs to be unencumbered by high radiation
levels,
Head Removal. Ordinarily, for example dur-
ing reactor refueling, removal of the reactor ves-
sel head is relatively straightforward. In the case
of Unit 2, however, the control rod drives that
penetrate the head may be entangled with the dam-
aged core. GPU Service Corporation plans to
place horoscopes 68 into the instrument or control
rod thimbles (penetrations made for control rods)
so that their condition can be evaluated visually.
If there is resistance when the control rods are
lifted, the drives will have to be disconnected prior
to removing the head. This task will involve, if
necessary, the removal of those head penetrations
which are not jammed first, in order to create
openings in the reactor head through which the
entangled drive-trams can be reached. The prob-
lem rods can then be cut loose and the head re-
moved.
All activities involving head and core removal
will be tested on mock-ups. As noted, special tools
will have to be designed and built, based on needs
to be defined as the cleanup proceeds. This type of
specialized tool design has been carried out suc-
cessfully during core removals at other reactors
which have experienced accidents. 89
Core Removal. The largest single source of
radioactivity is the reactor core, estimated to con-
tain 6 billion curies. Although it is generally easier
to control the spread of contamination from solids
such as the core than it is from liquids or gases, the
core at TMI has been severely damaged and poses
hazards that will require special handling.
The physical configuration of the core is un-
known. As a result of the accident, at least 90 per-
cent of the fuel rods have burst. (178) Periodic in-
jection of cold water into the extremely hot core
*"' The reactor head is bolted on the pressure vessel. It is massive but is designed to be removable.
* A horoscope is a device similar to a periscope, which allows remote viewing of objects. It has its own light source.
" See "Three Mile Island in Perspective : Other Nuclear Accidents," Append!* A, pp. 221-226.
187
(close to 4,000 F in some regions) probably caused
both the cladding and the reactor fuel itself to
shatter like glass. Some of the materials in the
core are thought to have melted. 70 This material
includes some of the silver-indium control rods as
well as the stainless steel in which these rods were
enclosed. These molten metals may have slumped
to the lower portions of the core and solidified
there, forming a casting of sorts.
It is difficult to determine with certainty what
the core looks like today. Some analyses suggest
the core resembles an empty bowl with fragmented
pieces of fuel and Zircaloy interspersed between
intact remnants of the fuel pins at the bottom of
the core. (179) The fuel assemblies further from
the center may be entirely intact, forming the
walls of the bowl. The upper portion of the fuel
in the radial center of the core was probably de-
stroyed and displaced, forming the cavity of the
bowl.
Great caution will have to be taken during core
removal to guarantee subcriticality. The boron
concentration must be maintained continuously at
3,500 parts per million until the core is out. While
the exact procedures that will be used are uncer-
tain, past experience points to some possible
approaches.
All handling and manipulation of the core will
be performed remotelv and under clean, borated
water, with the aid of underwater television. In-
tact fuel assemblies, damaged assemblies and loose
debris will be encased in metal cans underwater
and their ends welded shut with underwater weld-
ing equipment. The cans will then be moved
through the fuel transfer port to be placed in
either a shipping or storage cask. Several inde-
pendent neutron monitors will be put in place to
detect any increase in neutron activity, a sign of
recriticality.
With respect to the cost of fuel removal, the
July 1970 Bechtel estimate for fuel removal equip-
ment and disposal was $23 million. (180)
Worker radiation dose rates during core re-
moval are expected to be comparable to those en-
countered during normal plant refueling. 71 (181)
Since more workers will be exposed over a longer
time, however, the collective dose to the work
force will probably be greater than for refueling.
Nonetheless, the dose for this phase is expected
to be less than during decontamination of the con-
tainment.
Removal of the core is the last step in cleanup.
At this point the full degree of damage can be as-
sessed and a decision made as to the plant's future.
FUTURE OPTIONS
Four options have primarily been studied in
connection with TMI-2's future. 72 (182) This sec-
tion discusses some of the technical factors. 73
Decommissioning retiring the plant perma-
nently is a step normally taken after 30^0 years,
the projected life of a similar plant. If, however,
the facility can be reused, the utility has several
options including: to repair the existing nuclear
unit ; to replace it with a new nuclear unit ; or to
convert it to a coal-fired or other fossil-fueled
unit. 74
Obviously, the decision will depend in part on
the condition of both the nuclear steam supply
system (NSSS) 75 and the rest of the plant. While
rough estimates of the extent of the damage to
the NSSS and the plant exist, an accurate assess-
ment will not be possible until cleanup is complete.
At that time, radiation levels should allow rela-
tively unrestricted access throughout the contain-
ment, and a detailed evaluation can be made.
The technical decision will also depend on other
factors, most particularly finances. In considering
the financial factor, it should be noted that some
basic costs pertain to all options. As GPU's Presi-
dent, Herman Dieckamp, noted in response to
questioning by the Subcommittee:
. . . cleaning up the plant . . . has a cost
associated with it of at least $200 million
out of that estimated $320 million for the
total . . . costs. So that is there irrespec-
tive of return to service. (184)
As of April 1980, GPU maintained that damage
to the facility was within a range that would per-
mit the plant to be recommissioned either with a
nuclear or a coal-fired steam supply system. (185)
Bechtel had estimated the physical effects of the
accident in its report, issued July 1979. (186) The
report stated :
Excluding the conditions that existed
during the hydrogen detonation, the
physical effects of elevated containment
pressures and temperatures during the
""The presence of any previously molten materials can only be guessed at until the core can be viewed.
" During refueling, many workers are inside the containment. The resulting dose to the work force is therefore
higher than in normal plant operations.
72 This section is intended to provide an idea of what the final stage of recovery disposition of the facility entails
and generally what options have been considered most seriously.
78 See subsequent subsections for details on financial, social and legal and regulatory factors.
"Gilbert and Associates performed a study for GPU Service Corporation that included cost estimates for natural
gas-fired units, but noted that "Coal would be the primary fuel due to current and proposed restrictions on oil, and the
unknown availability of natural or synthetic gas in the quantities needed for the installation." (183)
75 The reactor, steam generator and primary system piping, etc.
188
March 28, TMI-2 accident on the contain-
ment structure, systems and components
were probably minimal.
In localize'd areas, the possibility of
some instrumentation damage, hydraulic
snubbers leaking, grease fittings/lubri-
cated fittings dripping oil, etc., does exist-
The pressure and temperature . . . con-
ditions that existed . . . do not appear to
be detrimental to the equipment in the
containment for the short time period in
question. (187)
The extent of damage done to the primary sys-
tem by thermal shock had not yet been deter-
mined" Similarly, the extent to which hydriding 76
of the steel had led to embrittlement was not
known. These factors can only be known after
entry of the containment.
There is also the possibility further damage to
the plant has occurred or will occur subsequent to
the accident. It includes radiation damage to the
insulation on the electrical wiring and corrosion
of components submerged in the pool of water.
However, if the damage is relatively limited, as
anticipated by Bechtel. replacement of many of
the major components within the plant will not
be necessary.
DECOMMISSIONING AND REPLACEMENT
Adequate experience is available on decommis-
sioning: plants. (188) Studies on the costs of de-
commissioning: plants of comparable size, at which
accidents have not occurred, are about $30 million.
(189) The studv prepared for The President's
Commission on the Accident at Three Mile Island
put the cost of decommissioning TMI-2 at $192
million, with a range between $157-$241 million.
(lOO 1 ) According to the studv, about half the ex-
penditure would be associated with the disassembly
and removal of structures."
If the plant is decommissioned and retired, the
utility would still have the option of replacing
Unit 2 with a new facility. The cost of a nuclear
unit, including interim replacement power costs.
is estimated at between S2.3 and $3.1 billion. (192)
The comparable cost of a coal-fired plant is
estimated at between $1.9 and $2.6 billion, in 1979
fixed dollars, reflecting the lower construction cost
of that type of plant.
RECOMMISSIONING
GPU has stated that it would prefer to put Unit
2 back into operation using fully the undamaged
portions of the facility, but has not decided
whether a nuclear or coal-fired unit will be select-
ed. (193) Its investment in TMI-2 was over $1
billion, and the plant had been in commercial oper-
ation for only 1 year at the time of the accident.
Decommissioning would mean retirement of an
essentially new facility. Thus, the utility has an
economic incentive to reuse the plant.
The cost of repairing the unit and returning it
to nuclear service was estimated by consultants to
The President's Commission at between $1.0 and
$1.9 billion, with construction running from 45 to
69 months (the figure includes the costs of replace-
ment power). (194) The higher figure is based on
replacement of the entire nuclear steam supply
system.
GPU has looked into converting the plant to a
coal-fired unit. Dieckamp explained the utility's
conclusions during the November 8 hearing :
. . . with respect to the study of alterna-
tives to returning it as a nuclear plant, we
have felt that ... it was going to be im-
portant for us to have good solid detailed
studies that had, indeed, evaluated the
options. And so. we have been looking at
an option that would convert the plant to
coal firing.
* * *
It becomes possibly a very complex un-
wieldy configuration, and perhaps, not a
very productive plan. (195)
Insofar as environmental constraints and costs
were concerned,
TVe are . . . looking at ... the capacity
of that local air basin to handle coal fir-
ing, and in addition, there are the prob-
lems of handling ash and scrubber
sludge . . .
* * *
I think it's probably true that the incre-
mental cost to get the next 900 megawatts
of power is probably less if one reconverts
or maintains it as a nuclear plant. (196)
Gilbert and Associates, Inc. studied the coal
conversion option 78 for the GPU Service Corpora-
tion. (198) It estimated the cost at between $0.7
billion and $1.0 billion, with a construction time of
* Hydriding is similar to oxidizing : it is a chemical reaction at the surface of the steel. It causes embrittlement of
the steel.
" The study cautions that ". . . our assessment should he interpreted with the caveat that the present assessments
may change as better data become available in the future." (191)
a The report contains the following caveat : "Moreover, the numbers presented in the report are subject to major
changes in the course of the detailed analysis which is expected to be made in the Phase II study. Consequently, no
judgement as to technical or economic feasibility can be reached on the basis of this report." (197)
189
from 42 to 48 months. 79 Their estimate did not in-
clude energy replacement costs, the expense of off-
site sulphur dioxide removal equipment and facili-
ties, decommissioning and other items.
According to the various studies, repairing the
nuclear plant is the least costly, followed by re-
placement with a coal-fired plant, followed by re-
placement with a nuclear plant. These figures are
summarized in Table 3. 80
TABLE 3. Cost in billions of dollars
Option
Repair or replace nuclear reactor in existing
facility $1.0-$1.9
Replace nuclear reactor with coal in existing
facility $0.7-$1.0*
Decommission existing facility, build new coal
plant $1.9-$2.6
Decommission existing facility, build new nu-
clear plant $2.3-$3.1
Excludes large energy replacement costs.
FINANCIAL ASPECTS OF RECOVERY
Kecovery has raised a number of financial ques-
tions, such as who will pay for cleanup, the finan-
cial condition of the GPU companies, the possibil-
ity of bankruptcy and the effect of bankruptcy on
cleanup. More than 1 year after the accident,
there were still no clearcut answers to any of the
issues.
THE PROBLEM OF CASH FLOW
The two nuclear facilities at Three Mile Island
are 50 percent owned by Met Ed, 25 percent by the
Pennsylvania Electric Company (PENELEC),
and 25 percent by Jersey Central Power and Light
Company (Jersey Central), three utilities which
are in turn owned by GPU. 81 (200) The NRG li-
censes authorize Met Ed to operate the two units
and to receive, possess and use special nuclear
material for that purpose. (201)
The major electrical generation transmission
and distribution facilities of the three utilities
are physically interconnected, and the GPU com-
panies operate as a single, integrated electric util-
ity system. Thus, the energy generated at TMI-1
and TMI-2 before the accident was distributed
throughout the GPU system. (202)
Since the accident, the GPU companies have
been facing serious financial problems, despite sub-
stantial assets. 82 The financial problems are re-
flected in the declining value of the parent com-
pany's common stock. Before the accident, it was
selling on the New York Stock Exchange for more
than $17 a share (205) ; in mid-May 1980, the price
was between $5 and $6. (206) Bond ratings of the
three GPU utilities also have fallen. Before the
accident, Moody's Investor Service had given Met
Ed and PENELEC bonds an "A" rating, Jersey
Central bonds a "Baa" bond rating (207) ; in
March 1980, PENELEC and Jersey Central's rat-
ing had dropped to "Ba," Met Ed's even further
to"B." (208)
The GPU companies' principal financial prob-
lem has been cash flow. (209) Their problem was
created by the need to pay for cleanup costs and to
purchase electric power to compensate for the lost
output of Units 1 and 2.
In early 1980, GPU's working estimate was that
decontamination of Unit 2 83 would cost at least
$200 million, and there were indicators that revised
cost estimates would be far higher. 84 (211) One
management consulting firm predicted that total
cleanup costs could be "half a billion dollars or
much more." (212)
The major ongoing expense, however, has been
replacement power. GPU reported that during
1979, the utilities' costs ran from $20 million to
over $35 million each month, (213) figures that
may increase.
The GPU companies' $4.9 billion in total assets
do not provide a simple solution to the companies'
"The President's Commission did not estimate the cost of rebuilding Unit 2 as a coal facility. The estimates of
schedule and cost prepared by Gilbert and Associates are not directly comparable to those of the President's Commis-
sion, since different assumptions were used in formulating the results. Gilbert and Associates also estimated the costs
of conversion to natural gas. This option was found to be less desirable. The costs ranged from $446 million to $500
million. (199)
* The figures in this table are all approximate and do not reflect comprehensive cost analyses.
" See "Prior to the Accident," pp. 50-51.
82 For 1979, GPU and its three utility subsidiaries reported more than $4.9 billion in consolidated assets, an increase
of more than $300 million over 1978. Of these, Met Ed accounted for about $1.3 billion, PENELEC about $1.5 billion and
Jersey Central about $2.1 billion. (203) The GPU system is the 14th largest investor-owned utility system in the nation
in terms of both assets and revenues. (204)
8J As discussed earlier, cleanup is only the first step in dealing with Unit 2. GPU still must decide whether to
rehabilitate, convert or decommission the facility. Each option will entail additional costs. See "Technical Aspects of
Recovery," pp. 188, 189.
84 Pursuant to agreement, Met Ed. PENELEC and Jersey Central are jointly responsible for all operating and
maintenance costs associated with TMI. including those related to the accident, in the same proportion as their owner-
ship shares 50 percent, 25 percent and 25 percent respectively. (210)
190
cash flow problems. GPU's only significant assets
are the common stocks of its three subsidiaries;
(214) their assets in turn consist mostly of facili-
ties and equipment and are not liquid. (215) Fur-
ther, according to GPU's Treasurer, John G.
Graham, neither GPU nor its utility subsidiaries
have had large cash reserves to draw upon in order
to help pav the post-accident costs. 85 (219)
The GPU utilities are paying for the accident-
related costs from three principal sources insur-
ance, loans and utility operating revenues.
The companies had $300 million in property
damage insurance for Unit 2, the maximum cover-
age available, and they expect the full amount to
be available to pay for cleanup. (220)
As of May 1980, cleanup costs were well below
the maximum coverage. 8 * However, insurance set-
tlements sometimes had lagged behind expendi-
tures, contributing somewhat to the cash flow
problem. ST
The GPU companies had no insurance to cover
additional replacement power costs. 88 They have
been trying to meet this major expense out of their
operating revenues and have requested rate relief
from State regulatory authorities in order to aug-
ment that revenue. 89
In June 1979. both the Pennsylvania Public
Utility Commission (PUCK which regulates Met
Ed and PENELEC. and the New Jersey Board
of Public Utilities (New Jersey Utilities Board),
which regulates Jersey Central, granted sub-
stantial rate increases to cover replacement power
costs. Nevertheless, for at least a year after the
accident, the rate increases did not match these
utility costs. 90 (226) This was because State
regulators had granted less than full relief and
because GPU's cost estimates proved too low. 91
In April and May 1980, New Jersey and Penn-
sylvania regulators granted substantial additional
rate relief designed finally to provide full and cur-
rent recovery of replacement power costs. 92
Although rate increases for replacement power
have been substantial, they have been offset to
some extent by other decisions by the Penn-
sylvania and New Jersey regulators. Most promi-
nently, the utility regulators have removed from
the utilities' rate bases all capital and operating
costs associated with TMI-2 and TMI-1. 93
Since insurance and operating revenues have
been insufficient to meet cash needs, the GPU com-
panies arranged to borrow money to help cover the
cash flow gap between existing expenses and fu-
ture revenues. They have obtained short-term
loans from a consortium of banks and some long-
term financing through institutional investors.
On June 15, 1979, GPU and its subsidiaries
entered into a revolving credit agreement with 45
banks, including Citibank, N.A. and Chemical
15 According to consultants working for the President's Commission on the Accident at Three Mile Island, the GPU
companies' financial structure prior to the accident was not unique. The consultants concluded that :
GPU [and its subsidiaries] followed general industry practices and. after reconciling individual company dif-
ferences, probably was not materially different in its financing practices, and results achieved, from the other
electric utilities which had facilities in New Jersey or Pennsylvania. (216)
According to GPU's Graham.
The virtually universal pattern for major electric utilities in the U.S. that own, as do the GPU companies,
generating, transmission and distribution facilities is to maintain virtually no balances of unrestricted cash
working capital . . . cash balances are generally only those required to be maintained as compensating bal-
ances for lines of credit and/or for outstanding short-term borrowings. (217)
In part, he said, this is because
The rate regulatory process has not permitted and is not intended to permit an electric utility to accumulate
large cash reserves to deal with the aftermath of an accident. (218)
"According to an NRC staff study, if Unit 2 is decommissioned, the property damage insurance will cover cleanup
costs, but not the costs of decommissioning, such as dismantlement or entombment. If the unit is restored to service,
restoration would be covered by any insurance remaining (up to the $300 million limit) after decontamination. (221)
GPU's estimates indicate cleanup and restoration costs may greatly exceed $400 million. (222)
87 According to GPU's treasurer, through October 31, 1979. approximately $83 million had been expended for con-
rainiug and cleaning up the accident, while insurance recoveries to that date were only $20 million. (223)
* Since the accident, members of the nuclear industry have formed an insurance pool through a mutual insurance
entity called Nuclear Electric Insurance Limited (NEIL) to help cover replacement power costs in the event of another
nuclear accident. Membership in NEIL will be available to electric utilities, including publicly-owned utilities, that have
an incurable interest in a nuclear power generating unit or in a nuclear unit's output. (224)
* See. generally. "Legal and Regulatory Aspects of Recovery," pp. 212-216, for a discussion of State regulatory pro-
ceedings since the accident.
* In December 1979. GPU's treasurer said that the "net outflow for replacement power purchases was running at
the rate of approximately $12 million per month." (225)
" In June 1979. Pennsylvania and New Jersey each approved rate relief that covered only about 85 percent of the
estimated replacement power costs of the three utilities. (227) The June 1979 rate increases had been based on utility
estimates of replacement power costs over 18 months. (228) Since the utilities had assumed that TMI-1 would be back in
service by January 1980. (229) their estimates of need over the 18-month period proved unduly optimistic. (230) The
companies also underestimated price increases in oil. which adversely affected the costs of purchased power. (231)
'-" In early June 1980. GPU said that it had spent about $300 million to replace lost output at TMI but had received
only $150 million of the increased costs from customers. (232)
" In June 1979. the Pennsylvania PUC and the New Jersey Board of Public Utilities each removed the Unit 2 costs
but permitted retention of Unit 1 costs. In April 1980, New Jersey removed Unit 1 costs, and in May 1980, Pennsylvania
did likewise. (233)
191
Bank. 94 The agreement established a line of credit,
initially totalling $412 million, with sublimits for
each GPU company. (235) As of late May 1980,
total borrowing could not exceed $292 million
without a favorable vote of the banks represent-
ing 85 percent of the credit line. (236)
The banks required that GPU put up substan-
tial collateral to back up the loans. For example,
GPU had to put up all of the common stock of its
subsidiaries. According to a Citibank official, be-
fore the accident, the banks would have given
GPU revolving credit without requiring
collateral. (237)
The agreement authorizes the banks to suspend
further credit, declare a default and accelerate the
due date of outstanding loans if, in the opinion of
the majority, there is a "material and adverse"
change in the actual or prospective financial con-
dition of the borrower that "substantially in-
creases the risk" that the loans will not be repaid
when due. (238) If the majority determines that
"the revenues to be available" to a borrower "will
be insufficient to assure its ongoing financial via-
bility," they may suspend further credit, although
they may not, for this reason alone, declare a de-
fault or accelerate the due date. 95 (240)
In late May 1980, after favorable ra*e rulings in
Pennsylvania and New Jersey, GPU officials
estimated that the loans would not reach the $292
million credit limit before the end of 1980, at which
time the banks would have to vote whether to in-
crease the credit limit. (241)
In late June 1979, PENELEC and Jersey Cen-
tral each issued $50 million worth of first mort-
gage bonds in order to obtain some long-term fi-
nancing. The bonds were sold through private
placement to a group of major insurance com-
panies. In each case, the net proceeds were to be
used either to pay outstanding short-term bank
loans or construction expenditures or to reimburse
the bond issuer's corporate treasuries for funds
previously expended. (242) In October 1979, Jer-
sey Central issued another $47.5 million in first
mortgage bonds under similar terms. 96 (244)
REGULATORY ISSUES
One effect of the stipulations on short- and long-
term borrowing has been to tie continued lending
to the rate-rulings of State regulators. During
Subcommittee hearings on November 9, 1979,
GPU's John Graham testified :
I would say if we receive adequate and
timely rate relief, I believe that Metro-
politan Edison Co. and the other two
operating companies of GPU will remain
financially viable.
* * *
With that hypothetical, favorable action
by the Pennsylvania Commission, I be-
lieve that the banks will stay with us;
Metropolitan Edison will have access to
bank credit; GPU can contribute in the
form of leaving retained earnings in
Metropolitan Edison Co. or by GPU
using part of the revolving credit agree-
ment to make borrowings at the GPU
level and to put that money into Metro-
politan Edison Co. as that might be nec-
essary. (245)
The lending banks have given similar testimony.
Officials of Citibank and Chemical Bank told the
Pennsylvania Public Utility Commission in Feb-
ruary 1980 that the PUC's decisions would "collec-
tively determine" the lending banks' confidence in
the GPU companies as viable entities to whom
continued credit should be extended. (246)
The Securities and Exchange Commission
(SEC) has been following GPU's borrowing ar-
rangements closely, pursuant to its statutory
duties. 97 (248) It, too, has pointed out the impor-
tance of rate relief. On November 9, 1979, SEC
Commissioner Loomis testified before the Sub-
committee :
Senator HART. Does the data you have
just presented indicate to you GPU is in
sound financial condition overall?
M Before the accident, the GPU companies had informal lines of credit totalling $225 million at 80 diffrent banks.
The reason for a written revolving credit agreement, according to bank officials, was that under the informal arrange-
ment any one bank could have unilaterally ceased to fund its line of credit, which would have left the other banks
"deeply concerned" and could have created a "cascading effect," with one cancellation leading to cancellations by the
others. The revolving credit agreement was designed to prevent this by setting up written procedures to insure the
lending banks acted "in concert." (234)
* The GPU companies have continued to maintain some informal credit lines with various banks. However, under
the revolving credit agreement, the amount of debt outstanding under these external lines cannot exceed $15 million.
(239)
"All bonds had a 20-year maturity; PEXELEC's bore interest at 11% percent per year, Jersey Central's 12 per-
cent per year (June issuance) and 11% percent per year (October issuance).
All of these bonds are subject to mandatory redemption in specified situations ; that is, they must be repurchased
by the issuer at face value prior to the maturity date. One situation would be if a bank participating in the revolving
credit agreement refuses to make advances to the utility. (243)
97 GPU is a holding company subject to regulation under the Public Utility Holding Company Act of 1935, 15 U.S.C.
79 et seq. According to SEC Commissioner Philip A. Loomis, Jr., one of the Act's purposes is ". . . to require that
the members of holding company systems be soundly financed without too high a level of debt . . . By requiring proper
financing, the act and regulations seek to keep each utility company highly solvent and most unlikely candidates for
bankruptcy, except under extraordinary conditions, which may . . . [exist] here." (247)
192
LOOKIS I will make an introductory
answer. Its capitalization is appropriate
for its operations, we believe, and it seems
to me ... the particular problem of GP
and Metropolitan Edison is the fact that
with both of these nuclear plants down
and not operating, they have to buy elec-
tricity from other sources, and that elec-
tricity is very expensive, and the problem
is whether or not their revenues will carry
the cost of obtaining purchased power.
(249)
The Director of the SEC's Division of Corporate
Regulation, Aaron Levy, also testified on this issue.
Senator HART. . . . assuming the re-
placement power costs will be covered by
the rate base for the utility, in your judg-
ment, should GPU be able to sustain the
estimated $300 to $400 million cost of the
TMI cleanup ...
LENT. On the basis of the ... [data]
we have, and if adequate rate relief is
granted. I can't see any reason why the
system should not be able to absorb what-
ever the cleanup costs may be. (250)
The regulators in Pennsylvania and New Jersey
have indicated their awareness of the link between
their rate-making decisions and the financial con-
dition of the GPU companies. (251) Their deci-
sions have reflected an attempt to provide the
utilities with needed rate relief without making
customers bear an unreasonable or inequitable
share of the utilities' costs. (252)
Thus, for example, on April 1, 1980, the New
Jersey Board of Public Utilities rendered a deci-
sion in which it said :
It is obvious that this availability of
funds [from the lending banks] is uncer-
tain and contingent on the banks' reaction
to regulatory action taken in Xew Jersey
and Pennsylvania. (253)
It also said :
The Board [of Public Utilities] clearly
recognizes the serious financial condition
of ... [Jersey Central]. This Board will
endeavor to work toward [preserving
Jersey Central] ... as an ongoing con-
cern to avoid the potential devastating
impact of insolvency or bankruptcy. 98
(254)
As of late May, rate adjustments by the New
Jersey Utilities Board had raised Jersey Central's
rate levels by some $293 million, roughly 48 percent
over the utility's pre-accident levels. About $82
million of the" increase was attributed to TMI.
(257)
In May 1980 the Pennsylvania PUC granted
PENELEC and Met Ed significant rate in-
creases, stating that their decision provided "an
adequate framework for Met Ed's recovery" and
that it was Met Ed's burden to "convince its bank
creditors that it ... has the will and the ability to
rehabilitate itself." (258)
At the end of May, rate adjustments by the
Pennsylvania PUC had raised Met Ed's rate levels
by some $150 million, 50 percent over pre-accident
levels ; " about $143 million was attributed to TMI.
(260) The rate levels for PENELEC, which did
not have to make substantial post-accident pur-
chases of replacement power had increased about
$19 million, 4.2 percent over its pre-accident levels.
All of this increase was attributed to TMI. (261)
Despite the utility regulators' actions, it re-
mained uncertain in late May how far the efforts
of the utility regulators would go toward resolving
the GPU companies' financial difficulties. 100
** Shortly before this decision. Coopers & Lybrand, an accounting firm, had submitted a report on the 1979 consolidated
balance sheets and related consolidated statements of GPU and its subsidiaries. The auditors warned that
The Corporation [GPU] is unable to determine the consequences of the accident . . . and of the response of
rate-making and other regulatory agencies to that accident
* * *
The Corporation's subsidiaries are currently not receiving a level of revenues sufficient to assure their ability
to continue as a going concern. The continuation of the Corporation as a going concern is dependent upon ob-
taining adequate and timely rate relief and maintaining and increasing the availability of credit under the
revolving credit agreement . . . The eventual outcome and effect of the foregoing on the consolidated financial
statements cannot presently be determined. (255)
Similar warnings were contained in the separate reports prepared for two of the three subsidiary utilities, Met
Ed and Jersey Central. The report for PENELEC. however, did not include the latter, more serious, warning quoted
above. (2561
* At the time of the accident. Met Ed had received approval for but had not yet implemented certain rate increases.
In the aftermath of the accident, these increases never went into effect. Assuming these increases had been implemented
at the time of accident. Met Ed's percentage increase in rates between that time and late May 1980 would have been
30.3 percent rather than 50 percent. (259)
"* On May 15. 19SO. the banks wrote to GPU saying that rate rulings in April and May had been "significantly respon-
sive" to many of the borrowers' needs but that "substantial questions" remained as to the borrowers' ongoing financial
viability. The bank indicated particular concern over the removal of Unit 1 costs from Met Ed's rate base. (262) See
"Legal and Regulatory Aspects of Recovery," p. 216, for further details.
193
NRC PROCEEDINGS ON TMI-1
Proceedings are underway before the NRC to
determine whether Unit 1 may be returned to serv-
ice. 101 If Unit 1 is returned to service, replacement
power costs may drop an estimated $14 million per
month. (263) Moreover, the restart of Unit 1 could
well cause the Pennsylvania and New Jersey util-
ity regulators to return Unit 1 capital and operat-
ing costs to the utilities' rate bases. (264) These
costs had accounted for about $56.5 million in
annual revenues to the three utilities. (265) If Unit
1 cannot be brought back to service, the GPU com-
panies will finally have to decide how permanently
to replace Unit 1's energy output at a reasonable
cost.
As of late May, formal hearings were not ex-
pected to begin before the fall, and there was no
firm date for a final decision. (266)
THE PROSPECTS OF BANKRUPTCY
At the Subcommittee's November 1979 hearings,
GPU's Treasurer Graham stressed that he had "no
reason to believe that Met Ed will become insol-
vent." 102 (268) Nor did GPU's President Dieck-
amp "see the situation where we perceive to have
bankruptcy of Met Ed to be in the best interest of
GPU." (269) Graham also has stated that bank-
ruptcy would be "seriously adverse" to "investors
in the securities of the GPU companies." (270)
The lending banks have indicated that bank-
ruptcy of the GPU companies would not neces-
sarily be in their best interest either. According to
February 13, 1980 testimony of a Citibank official :
From our background, what we do know
of [utility bankruptcy] we feel very
strongly it's not likely, as we've stated in
our testimony, to benefit anybody but the
legal profession. (271)
These statements provide some reassurance, but
business considerations change, and if one or more
of the GPU companies fails to pay its debts as they
come due, bankruptcy cannot be ruled out.
THE EFFECT ON CLEANUP
A corporation does not necessarily close down
and liquidate its assets under bankruptcy law. The
company may instead go through a judicially-
supervised reorganization while it continues to do
business. 103 (272)
SEC Commissioner Loomis advised the Subcom-
mittee that "as a practical matter" a utility com-
pany such as Met Ed, which is providing electric
power to the public, could not "simply close down
and turn off the lights and liquidate its as-
sets." 10 \ (274)
Loomis also testified, however, that in the event
of Met Ed's bankruptcy there was no assurance the
utility's revenues would be directed to cleanup :
. . . [Bankruptcy] would raise some very
difficult, unsettled questions under the
new Bankruptcy Act as to who would get
whatever revenue comes in; whether it
would go to cleanup or whether it goes to *
paying off bonds is unsettled under the
present law. Though this is a brand-new
law, I think the courts would decide it
right, but there would be a lot of litiga-
tion. (275)
GPU's Treasurer gave similar warnings about
the uncertainties of bankruptcy, noting that credi-
tors would likely argue against the use of utility
revenues for purchasing replacement power or for
prosecution of the TMI-2 cleanup efforts. (276)
The question of Met Ed's bankruptcy raises the
issue of what responsibility the remaining GPU
companies would have for cleanup.
According to a memorandum on a staff inter-
view, GPU^ Treasurer had "committed GPU to
cleaning up TMI-2 no matter what circumstances
transpired." (277) When asked about this at Sub-
committee hearings, Graham replied :
As I recall the context in which we were
discussing the issue with your staff, we
were talking about the three operating
companies working together outside of a
bankruptcy or an insolvency of one of
those three companies, and I was saying,
and I continue to say, we would take all
steps that we can and within our power to
do the cleanup job. . . . (278)
During the Subcommittee hearings, GPU Presi-
dent Dieckamp similarly indicated that because of
the "many uncertainties," he could not unequivo-
cally commit GPU's resources to cleanup. (279)
101 See "Legal and Regulatory Aspects of Recovery," p. 212, for a more detailed discussion of this proceeding.
103 A debtor corporation may voluntarily commence bankruptcy proceedings. Involuntary bankruptcy proceedings
may be instituted against a debtor corporation by creditors. (267)
""Under 15 U.S.C.A. 79k (f) (Supp. 1980), If an action were commenced in Federal court for bankruptcy reorga-
nization, the SEC would have the right to be heard concerning the appointment of a receiver or trustee for a registered
holding company such as GPU, and the SEC could itself be appointed. Any reorganization plan would have to be ap-
proved by the SEC before it could become effective. Even without commencement of a bankruptcy proceeding, a holding
company like GPU may submit a reorganization plan to the SEC, pursuant to 15 U.S.C. 79k (e), and the SEC would
have the power to approve the plan and present it to a Federal court for enforcement.
104 If a bankruptcy liquidation proceeding were commenced, a party in interest could ask the court to convert it to a
reorganization proceeding. Moreover, the debtor would have a one-time absolute right to convert it to a reorganization.
(273)
194
In light of GPU's stated position, the Subcom-
mittee Chairman said :
... I think what our concern is if GPU
will not stand behind that obligation and
Met Ed does go into insolvency or receiv-
ership what entity is legally obligated to
maintain . . . that plant, and keep the core
cool . . . [W]e are trying to figure out here
whether it is Met Ed's responsibility.
GPU's responsibility, the NRC's respon-
sibility, the State of Pennsylvania's re-
sponsibility, the Congress' responsibility,
or whose responsibilities it is. And on the
answer to that question, in my judgment,
could well rest a large part of the future
of that industry. . . . (280)
Subsequent to the hearings. Graham sent the
Subcommittee a letter setting forth the "prelim-
inary results" of an investigation into whether
GPU. PENELEC and Jersey Central would be
liable for cleanup costs if Met Ed were to enter
bankruptcy reorganization proceedings. (281)
Without ever stating "no," he gave an assortment
of reasons why the three remaining GPU com-
panies might not be liable. 105 One suggestion was
that, at most, only Met Ed might be held legally
responsible for cleaning up the facility, even
though the three utilities jointly own the Unit 2
facility and pay for operating expenses.
THE FUTURE
At the time of the accident, the NRC did not
require that licensees have sufficient insurance or
other financial resources to deal with a nuclear
accident. 106 Xor did it require that a holding com-
pany's assets be legally committed to cover any
cleanup costs at a subsidiary's nuclear plant. (284)
When this point was raised during the Sub-
committee hearings, the NRC Commissioners ex-
pressed interest in the idea of ensuring that suffi-
cient funds be available for cleanup of an acci-
dent, but did not indicate that they had taken any
steps to that end. 107 Less than 3 weeks after the
hearings, the Commission directed the agency's
Executive Director for Operations to study al-
ternative approaches to assuring such arrange-
ments, including the possibility of requiring in-
surance coverage and "a commitment of a holding
company's assets for accident recovery." (286)
As suggested earlier, cleanup efforts by the GPU
companies would not necessarily stop in the event
of bankruptcy proceedings. Those proceedings pro-
vide a method for determining the bankrupt's
1(0 Graham's letter asserted that :
"(a) The Atomic Energy Act, as amended, does not on its face impose a clear statutory obligation on the owner or
a nuclear facility to clean upon the consequences of an accident and the regulations adopted by the XRC do not on their
face impose such an obligation. Similarly, the license for TMI-2 does not expressly impose such an obligation. While
the GPU companies do not dispute the existence of such an obligation and intend to meet it if permitted to do so, the
question as to whether such an obligation exists might have to be resolved in litigation if there were to be reorganiza-
tion proceedings.
"(b) Assuming that such an obligation exists, there is nothing in the Atomic Energy Act or regulations or license
which would cause such an obligation to be other than an unsecured general claim and, as such, subject to the prior
liens of the mortgage indentures securing the first mortgage bonds of the TMI-2 owners.
"(c) Although the licensees of TMI-2 are Met Ed. Jersey Central and PEXELEC, the license grants only Met Ed
the power to operate TMI-2 and to receive, possess and use special nuclear material for that purpose. Resolutions of
the potential questions referred to in subparagraphs (a) and (b) without litigation would appear to be even more
doubtful in the case of Jersey Central and PEXELEC.
"(d) PEXELEC and Jersey Central have an agreement with Met Ed whereby they have each agreed to provide
25 percent of the cost of operating and maintaining TMI-2, and Met Ed is to provide 50 percent of such costs. It is not
clear whether, if Met Ed were involved in bankruptcy proceedings, it could use its revenues (which have been pledged
to secure its bonds) or other mortgaged property to pay for its share of such costs. If Met Ed were able to provide for
its share of such costs and Jersey Central and PEXELEC were not themselves in reorganization, such contract would
call for pro rata payments by Jersey Central and PEXELEC. although counsel for one of the intervenors in the Jersey
Central proceedings has questioned the validity and effectiveness of that agreement. It may also be argued by someone
that the contract does not cover cleanup costs unless they are a part of maintenance costs incidental to restoring TMI-2
to service . . .
"(e) GPU is not a co-licensee under the operating license and does not have an agreement with Met Ed. As pre-
viously pointed out, GPU has virtually no assets other than the common stocks of Met Ed, PEXELEC and Jersey
Central." (282)
"* Section 182(a) of the Atomic Energy Act of 1954, as amended, 42 U.S.C. | 2232 (a), states that each license applicant
"shall specifically state such information as the Commission . . . may determine to be necessary to decide such of the
technical and financial qualifications of the applicant ... as the Commission may deem appropriate for the license."
Under 10 C.F.R. 50.33(f) and Appendix C. an applicant must demonstrate that it has reasonable assurance of
obtaining the funds to cover estimated operating costs and costs of permanently shutting the facility down. According
to XRC staff, this regulation has been interpreted to cover a normal decommissioning operation, not the more substantial
costs resulting from an accident. (283)
107 For example, former Commission Chairman Joseph Hendrie said :
I guess the financial side does have an interest here. You would want to have reasonable confidence that
you weren't licensing a plant or a utility that was in such shakey condition that they would just go into bank-
ruptcy and there would be some question about their survivability as an operating entity to take care of the
site. Yes, I think we have to look. I am not quite sure how we treat or how well you could do any analyses, but
I think we need to look at it. (285)
195
obligations and debts and for deciding which will
be satisfied. (287) Further, the NRC's General
Counsel said that if such action were necessary, the
agency has the authority to run the TMI facility,
pursuant to Section 186 (c) of the Atomic Energy
Act, 42 U.S.C. section 2236 (c) , 108 (288)
During the November 9 hearing, the Subcom-
mittee Chairman asked NRC's General Counsel^
Leonard Bickwit, about the possibility of an NRC
takeover :
Senator HART. If Metropolitan Edi-
son were to go into receivership or become
insolvent, and for one reason or another
GPU were unable or unwilling to assume
responsibility, what would be your rec-
ommendation to the Commission in this
regard ?
BICKWIT. Yon want to look at the op-
tions, but at this point I would advise
them, if it happened tomorrow, I would
advise them to take it over, and if the
expertise of the Commission was not up
to the problem, contract with those who
could assist. (289)
Following up on this latter point, the Subcom-
mittee asked the NRC's Denton whether NRC staff
was capable of taking over the plant. Denton
replied :
I think the answer is yes ... I do think
the NRC operation could assume a man-
agerial, technical direction of the plant,
but this is only an assumption that many
of the employees of the plant who are
skilled in operating individual pieces of
equipment could be transferred and some-
how paid by the NRC. We don't have
the operational capability to replace
those individual employees that are ac-
tually manning the equipment today.
And to do that, would require a mas-
sive rearrangement of our own priorities
and assistance from other Government
agencies. (290)
At the hearings, the Subcommittee Chairman
also asked if the Federal Government would have
a responsibility to pay for the cleanup in the event
of bankruptcy. Bickwit replied, "I don't see it."
(291) Commisioner Hendrie agreed:
. . . unless we get to some situation where
it is an urgent public safety matter and
there simply isn't any other institution
around that is able to take action. But
short of that, which I don't see as being
the case, it is not a Federal responsibility,
in a financial sense, I wouldn't think.
(292)
In late February 1980, an NRC Special Task
Force on Cleanup noted that bankruptcy of a
licensee was a risk for which no contingency plans
had been prepared :
There is some risk that the licensee may
go bankrupt and may not be able to com-
plete the cleanup. There are no known
plans to cover this contingency. (293)
It recommended that the
Commission, in conjunction with other
government agencies, prepare contin-
fency plan for cleanup in case of financial
lilure of licensee. (294)
As a result of this recent recommendation, the
agency has begun to prepare contingency plans for
NRC management of the cleanup. (295)
IN SUMMARY
The financial aspects of cleanup involve the
weighing of many interests. The GPU companies'
financial condition has been largely tied to the
decisions of utility regulators, who have had to
balance the needs of the utilities and their custom-
ers. The NRC has had to fulfill its mandate to
protect the public's health and safety, yet its deci-
sions also may affect the financial condition of the
utilities. More than 1 year after the accident,
the financial condition of the GPU companies re-
mained uncertain, as did the consequences of bank-
ruptcy on cleanup.
SOCIAL ISSUES IN RECOVERY
The accident at Three Mile Island has remained
a major source of anxiety for local residents. The
President's Commission on Three Mile Island
concluded that "the most serious health effect of
the accident was severe mental stress." (296)
Strong distrust and lack of confidence in the NRC
and the licensee have persisted during recovery.
On November 8, 1979, the Subcommittee heard
testimony from two elected officials from commu-
nities near TMI. One was Bruce Smith, Chairman
108 Section 186(c) , 42 U.S.C. section 2236(c) , states,
In cases found by the Commission to be of extreme importance to the health and safety of the public,
the Commission may recapture any special nuclear material held by the licensee or may enter upon and operate
the facility prior to any of the procedures provided under the Administrative Procedure Act.
196
of the Board of Supervisors, Newberry Township,
a township located just a few miles from TMI.
(297) He described himself as "an average citi-
zen" and a "conservative,'' who before the accident
would "invariably compare the cooling towers to
the pyramids." (298) He stated :
. . . [N]ow I am so angry about Three
Mile Island that I have become one of the
leaders in the movement to close TMI
forever, as a nuclear plant. (299)
Smith cited a specific example of the commu-
nity's ongoing distrust. Referring to Met Ed's No-
vember 1979 request to vent the krypton gas in the
containment, 109 he testified :
I personally attended the news conference
when Met 'Ed announced their desire to
release krypton into the atmosphere. Met
Ed officials seemed mystified when local
citizens protested; after all the krypton
only had half-life of [a little] . . . more
than 10 years. It was little consolation to
the people of central Pennsylvania to
know that Met Ed was going to select the
days when wind direction and velocity
were best for release of the krypton. 11 '
(301)
Smith said further :
. . . [T]he bottom line of what most peo-
ple say is due to their unique experience,
they don't quite believe everything that
they're told ... So. the people don't
know what to believe, and they're told
that everything is being done safely and
within the guidelines and acceptable
limits. Even the word acceptable limits
becomes laughable when you've been
through what people in central Pennsyl-
vania feel they've been through. (302)
Smith recommended one way to improve the
community's attitude toward cleanup :
A long-range step-by-step plan could bet-
ter prepare the community as well as the
community leaders with the problems
and dangers to be confronted with the
cleanup process. (303)
Referring to some of the many post-accident
studies and surveys of local residents, Smith also
noted that the constant reminders of TMI might
be fueling public concern :
[T]he inherent problem is similar to that
of a hypochondriac who learns of too
many potential diseases. It becomes a
psychological problem which depresses
the interviewer and the interviewee . . .
The psychological impact of the accident
at Three Mile Island is immeasurable,
but it is there, in many homes. 111 (305)
A second Subcommittee witness was Albert B.
Wohlsen, then the Mayor of the city of Lancaster.
(306) As is discussed later, Lancaster city officials
had initiated a civil action in May 1979 to enjoin
the NRC from permitting the discharge of any
Unit 2 wastewater into the Susquehanna River, a
major source of Lancaster's drinking water. 112 The
reason for the suit, according to Wohlsen, was
that cleanup decisions "were being made with no
opportunity for Lancaster's participation." (307)
Wohlsen gave his view of community distrust :
"[t]he inaccuracies, inconsistencies and misinfor-
mation supplied by Met Ed and the Nuclear Reg-
ulatory Commission following the accident" had
produced in citizens from the Lancaster area a
"crisis of confidence concerning the ability of Met
Ed and the NRC to protect the public." (308) He
added,
Met Ed and the NRC have made repeated
assurances that their post-accident pro-
cedures are more reliable, accurate and
responsive to the public's need for reli-
able information. That conclusion, how-
ever, is open to serious challenge. (309)
Further,
. . . Restoring public confidence in nuclear
power and our governmental ability to
safety control it both in Lancaster
County and elsewhere, will require more
effort in the future than has been demon-
strated by Met Ed and the NRC in the
past. (310)
108 See "Legal and Regulatory Aspects of Recovery," pp. 205-207, and "Technical Aspects of Recovery," pp. 182-184, for
details on Met Ed's request to vent the krypton, the XRC's response and the events that followed.
110 Met Ed. in requesting XRC's permission to proceed with venting, addressed the problem of public concern :
We are cognizant of the concern on the part of some members of the surrounding communities about the venting
of the Kr-85. We are convinced, however, that this is the most prudent and safest approach . . . The Com-
pany will do whatever it can to provide sufficient information to the public to assure them they will be aware
of the timing of releases and the results of the monitoring on both on-site and off-site radiation levels. (300)
111 According to a newspaper report, psychologists consulting for the National Science Foundation recently
", . . found a direct relationship between the degree of risk perceived by laymen and the frequency with which
a potential risk is mentioned in news reports. During the Three Mile Island incident, for instance, some 700
newsmen, editors, photographers, producers and support staff were on the scene concentrated news coverage
matched in recent years only by Saigon during the height of the Vietnam war." (304)
112 See "Legal and Regulatory Aspects of Recovery," pp. 201-204, 207.
197
Wohlsen, too, suggested a solution :
The public must be fully involved and in-
formed so that it can be confident that re-
actor accidents are openly and properly
analyzed and resolved. (311)
A third Subcommittee witness was Judith
Johnsrud, Co-Director of the Environmental Coali-
tion Against Nuclear Power, a non-profit organi-
zation representing "individuals and citizen
groups throughout the Pennsylvania and adjoin-
ing states." (The Coalition had intervened in the
licensing proceedings involving TMI-2). (312)
Stating that area residents were "trying to restore
some semblance of sanity to their own lives," (313)
she commented on the "distressing lack of ...
reliable information [available] from official
sources." (314) She also said,
We find that, in all except the most out-
spoken proponents of nuclear power and
the most apathetic, there is a sense of un-
ease. Although many people appear to be
unwilling to discuss the persistent haz-
ards of the plant, when pressed they ad-
mit they are sick of the matter and just
wish the problem would go away. There
is little sympathy expressed for Met Ed ;
we find few who believe in either the
veracity or competence of the utility to
conduct the recovery or further operation
of the reactors at TMI. (315)
During the Subcommittee's hearings, GPU's
President Dieckamp acknowledged the continuing
concerns of the local public :
But, the cleanup is more than a technical
matter. It involves activities which have
been perceived by the local public as im-
posing an unknown hazard. The accident
has made some segments of the public so
conscious and fearful of radiation that
there's a great tendency to accept nothing.
(316)
He added that "we certainly recognize there's a
great need to inform the public and in the process,
to hopefully regain some public confidence." (317)
According to Harold Denton, the NRC also was
"acutely aware of the need to keep the local citizens
and governments informed." (318) He suggested,
I think we can bring this to a much better
focus and lay out for the public inspection
general plans so that everyone can under-
stand what are the steps and still provide
flexibility for adjusting and modifying
the plan as new knowledge is gained.
(319)
At the time of the November Subcommittee
hearings, the NRC, Met Ed and the State of Penn-
sylvania were trying to restore community trust
by holding biweekly meetings open to the public.
(320) In addition, once the NRC decided to pre-
pare an environmental impact statement, 113 public
"scoping" meetings were held to discuss this doc-
ument. (321) By early February of this year, the
NRC also had set up a permanent office in Middle-
tow T n, Pennsylvania, both to serve as an offsite base
for NRC officials and to make the agency more
accessible to the public. (322)
Nonetheless, fueled in part by the accidental re-
leases on February 11, 12, and 13, 1980, 114 commu-
nity concern and distrust persisted. On Febru-
ary 12, 1980, the NRC held a public meeting near
TMI to solicit comments on the programmatic en-
vironmental impact statement being prepared on
the decontamination and disposal of radioactive
wastes at TMI. A woman who described herself as
a 36-year resident of Middletown said at that
meeting :
... I work in mental health . . . Now,
I am seeing among people I know, just lo-
cal people, my neighbors, the same kinds
of symptoms I am seeing in people I am
treating, only we accept it as normal. We
have come to a place, living here, where
we have accepted high anxiety, stress,
fear, and inability to sleep, restlessness,
the desire to escape, a feeling of being
trapped, we have begun to accept that as
normal. And that is not normal.
* * *
. . . people are really, are being impacted
on a daily basis by things that they are
beginning to believe they cannot in any
way change. That induces hopelessness.
Hopelessness induces depression. And if
we don't get cancer from radiation, then
the effect of depression will probably take
its toll. (324)
Some speakers stated that they could no longer
rely on the licensee and the NRC, and two citizens
suggested that a local citizens' advisory group
be funded to conduct an independent review of the
activities at TMI. (325) One commented :
. . . [T]here should be a citizens advisory
panel. I think it should be, that you
111 See "Legal and Regulatory Aspects of Recovery," p. 201, 204-205, for further discussion of the impact statement.
114 In a report dated February 28, 1980, an NRC Special Task Force on Three Mile Island Cleanup reported the view
of on-site NRC support staff :
. . . that there had been considerable improvement in the public's confidence in the licensee during the past
10 months, but that this confidence was severely eroded by the events that took place at TMI-2 and were so
widely publicized during the week of February 11, 1980. The Mayor of Middletown expressed a similar view.
(323)
198
should make a real, deliberate attempt to
contact some of the leaders of these local
organizations, who have been active, who
have tried to educate themselves, and who
have a tie with the community, know what
the people's concerns are. and deal with
them on a day-to-day basis.
I also think that that advisory panel
should have funding provided so that
thev can solicit input from qualified, in-
dependent experts to help evaluate these
assessments that you people are doing, so
that we feel that we are getting the input
and we are able to ask the questions and
get the type of information that we feel
good about. (326)
In its February 28, 1980, report to the Commis-
sioners, the XRCs Special Task Force on Three
Mile Island Cleanup noted that "[tjhere exist
strong feelings of fear and anxiety among citizens
about the activities at TMI-2." (327) According
to the Task Force :
The public concerns for health and safety
appear to stem from a lack of public con-
fidence in either the licensee or XRC.
coupled with a conviction on the part of a
substantial fraction of the population
that releases of any quantity are danger-
ous and or that the magnitude of releases
is consistently understated. These con-
cerns have led to a high degree of stress
for a segment of the population, which
needs to be alleviated. (328)
The Task Force recommended consideration of
a "citizen's advisory committee" in connnection
with the preparation of the environmental impact
statement. (329) Further,
Staff . . . [should] take positive actions to
ensure local citizens are (a) informed of
the need for timely cleanup of TMI and
the steps to be taken to clean up the plant,
including evaluation of alternatives; (b)
alerted when particular planned releases
are to be made, with advice on precautions
the public should take, if any: and (c)
provided data promptly about radiation
levels in their communities during the
course of any release. (330)
Several weeks after the Task Force issued its
report. XRC staff recommended to the Commis-
sion that it approve the "controlled purging" of
krypton gas in the containment. 115 In doing so.
the staff stated that it was
fully aware of the public sentiment
against the planned or accidental release
of any further radioactive materials ,
regardless of the dose consequences . . .
[T]he authorization of controlled purg-
ing will entail some public concern and
stress despite the absence of significant
radiological health effects. On the other
hand, if purging is not authorized . . . ,
based on past experience there will con-
tinue to be planned and unplanned small
gaseous releases incident to the activities
involved in maintaining the facility in
a safe status as well as continuous low
level releases from offgassing . . . Thus,
even if purging is authorized there will
still be a source of continued public con-
cern and stress , but the major source
of public concern will have been allevi-
ated. (331)
On March 19, 1980, the NRC held a public meet-
ing in Middletown to discuss the staffs assessment
that venting would have no significant adverse
impact on public health and safety and no
significant environmental impact. The meeting
was punctuated by frequent interruptions by the
audience. (332) One speaker from the audience
explained :
. . . people aren't very polite to you to-
night and I would be willing to bet that
on the whole with maybe a few excep-
tions, this is a pretty law abiding, polite
crowd usually.
But the thing is that when you push
people to the wall and when you threaten
people's lives, when you threaten their
children's lives they are not polite. They
are afraid and they are angry.
These people are pretty angry tonight.
... If that anger is so bad tonight when
we are just talking about venting kryp-
ton, what is going to happen if you make
that decision to do it ? (333)
Another resident, who lived 3 miles from the
plant, commented :
Met Ed's alleged concern for my safety
insults me. They rightly assume that I
don't want any equipment to malfunction
from lack of maintenance, or even relive
another reactor accident.
However, they assume that I would
therefore willingly accept the low level
and I don't know how low level radioac-
tive releases, a far lesser risk they say
than relying on other alternatives.
I -have been blitzed by their PR cam-
paigns and their charts and their fancy
numbers and their smiling assurances
m See "Technical Aspects of Recovery." pp. 182. 183. and "Legal and Regulatory Aspects of Recovery," PP- 2(6-207, for
further discussion of the venting issue.
199
that the levels of radiation to be vented
are within Federal safety limits. But who
knows if the Federal safety limits are
safe (334)
Two days later, on March 21, the NRG held
another meeting, this time in the Commissioners'
conference room in Washington, B.C. Among
those present were three Commissioners and seven
members of a "citizens' group on TMI cleanup."
The citizens, who described the earlier public
meetings on venting as "rowdy," (335) repeated
to the three Commissioners an attitude expressed
previously :
. . . distrust, absolute distrust for the
variety of authorities who have been hop-
ing to be in control in the matter of
TMI ; that includes Met Ed, that includes
the NRC. (336)
One citizen added, "we are beginning not even to
trust the [State of Pennsylvania's] Department of
Environmental Resources." ue (337)
The Commissioners were told of efforts to de-
velop a local citizens group. According to one of
the community residents :
We are talking about ... a citizens group
that can act as a buffer between the Com-
mission and the citizens so that this does
not deteriorate into something far worse
and . . . get out of control. (338)
She added :
... If you decide to have a citizens com-
mittee, we don't want the appointments
made by any politicians or any bureau-
cratic offices. We can submit a list of
names that I feel perhaps will meet with
the approval of most of the people in the
TMI area who feel that their best inter-
ests are being served. We don't want any
appointments coming from the Governor,
from Washington, or from anyplace else.
The credibility is gone. We now feel that
we have to get in control of our own lives
and I would appreciate anything that you
could do in that area. (339)
It should be stressed that not all area residents
have opposed the cleanup proposals, including the
venting of krypton. Newspaper articles in late
March quoted local citizens as criticizing vocal op-
ponents of venting as "hysterical" or "disgusting"
(340) and insisting that it was "time for the silent
majority to come forward." (341) One article
noted that Middletown's population appeared to
have risen, probably because of the influx of work-
ers helping the cleanup, thus "belying any notion
that this is an atomic gnost town." (342)
On May 12, 1980, local citizens expressed this
different perspective to the NRC Commissioners.
At that time the Executive Secretary of the Penn-
sylvania Holstein Association stated :
It seems to us that this venting is an im-
portant, safe, and reasonable step in a
]ob that must be done : the prompt clean-
up of TMI. And actually the sooner it is
done, the better not only for the members
of our association, but for the entire
Commonwealth of Pennsylvania.
We are concerned that further delay
could result in possible deterioration of
the containment building and cause un-
controlled venting.
Any health risks to our citizens and
possible economic loss to our business
community must be avoided. The agricul-
ture community in Pennsylvania could
not withstand the economic loss that
would follow uncontrolled venting. (343)
This attitude was repeated by another area resi-
dent, who added :
When is TMI going to be cleaned up?
This latter question is of particular im-
portance. With a flood or other national
disaster cleanup begins as soon as the
damage subsides. The man on the street
can do something and within a few
months things are pretty much back to
normal. The disaster may be nearly for-
gotten.
But over 1 year later Unit 2 is still not
cleaned up and we are constantly re-
minded of the accident and the fact that
it is still potentially dangerous. The long-
er it takes to get everything cleaned up
the longer the citizenry will be subject to
rumors, lies, and varying degrees of un-
certainty.
I feel as soon as TMI is cleaned up and
either shut down or reopened concerns
will begin to dissipate. However, these
concerns may take a long time to disap-
pear because people will continue to won-
der where or when or if it will happen
again. Fear of the unknown is a very real
fear. That is not to say that the people of
Middletown are in a constant state of
anxiety or panic but as long as the fore-
cast for TMI remains unknown there will
certainly be fears and concerns.
1
116 As mentioned in "Legal and Regulatory Aspects of Recovery," p. 206, fn. 125, the Pennsylvania Governor's Com-
mission on Three Mile Island had indicated in a February 1980 report that it would not oppose prompt venting provided
that dose levels were "acceptable." The Secretary of the Department of Environmental Resources was a member of the
Commission.
200
For this reason we ask you to arrive at a
decision concerning the venting of kryp-
ton as soon as possible. (344)
In late March, the Governor of Pennsylvania
responded to those who had expressed concern and
distrust over venting. He requested that the Union
of Concerned Scientists (ITS), an organization
opposed to nuclear power, do its own analysis
of the XR( ' staff's proposal. He explained that the
proposal had stirred -considerable anxiety" in the
area and that he wanted to ensure that the plan
was analyzed "by the bi-oadest range of experts,
and 'in the lioiw of assuring our people that
whatever course ultimately taken is. indeed, the
safest available." (-^:>)
On May 14. 1980. the ITS released its results.
A- discussed in more detail elsewhere. 117 the ITS
concluded that the XRC staff's proposal would not
have any significant adverse health effects but
recommended against the particular venting
method proposed localise of the anxiety it would
cause area residents. (346) Two days later, on
May 16. Governor Thornburgh told the NBC he
would support a decision to proceed promptly with
the XRC staff's venting proposal because of what
he termed a "broad based consensus" among vari-
ous experts, including ITS. that the proposal
would not have direct radiation-induced adverse
health effects. (347)
IN SUMMARY
For a variety of reasons, concern and anxiety
still exist among some members of the community
surrounding TMI. State and Federal officials and
the licensee all have acknowledged and sought to
relieve these concerns. They have not been success-
ful. The fundamental, continuing problem is a lack
of trust and confidence in those who bear responsi-
bility for ensuring that cleanup is accomplished
expeditions!}-, but with due regard for the health
and safety of the public.
LEGAL AND REGULATORY ASPECTS
OF RECOVERY
The judicial and regulator}- proceedings that
have followed from the accident are complex and
involve many parties, among them the licensee.
Met Ed. the XRC and numerous Federal and State
agencies, public officials, and private groups and
citizens. The proceedings are necessarily delibera-
tive and therefore affect the pace and nature of
cleanup: they alsx> will affect the ultimate cost of
the accident.
PACE AND NATURE OF CLEANUP
A difficult problem the licensee and the XRC
have been facing is how to decontaminate and dis-
pose of the radioactive solids, liquids and gases in
Unit 2. Technical solutions have been complicated
by legal and regulatory factors.
EPICOR-H
EPICOR-II is a water purification designed
specially for TMI." S On May 20. 1979. as its in-
stallation neared. officials of the city of Lancaster
went to Federal district court to enjoin the XRC
from
. . . approving or allowing (a) the con-
struction or operation of any decontami-
nation equipment or piping, and (b) the
decontamination of or discharging into
the Susquehanna River of any radio-
active waste water from . . . reactor Xo.
2. ... (348)
The city charged that the XRC had "proceeded
secretly to select and approve decontamination
plans"" and insisted that the plans be "fully ex-
amined and subjected to public review and com-
ment." (349)
Among its legal claims, the city alleged that
before the XRC could proceed, it was required to
prepare an environmental impact statement cover-
ing all plans to decontaminate Unit 2 and dispose
of the radioactive water. (350) An impact state-
ment is required under the National Environmen-
tal Policy Act of 1969 (XEPA). 42 U.S.C. sec-
tions 4321 et sfq.. for "major Federal action
significantly affecting the quality of the human
environment." It sets forth alternative approaches
to the proposed project and how each might affect
the environment. (351)
On Mav 2.">. as a result of the city's lawsuit, the
Commissioners directed XRC staff to prepare an
environmental assessment for EPICOR-II. (352)
L#ss elaborate than an environmental impact state-
ment, an assessment is supposed to
|"b]riefly provide sufficient evidence
and analysis for determining whether to
:i: See "Technical Aspects of Recovery." pp. 171. 183. and "Legal and Regulatory Aspects of Recovery," pp. 206-207.
m See "Technical Aspects of Recovery." pp. 179-182.
201
-
prepare an environmental impact state-
ment . . . (353)
Ordinarily, the more detailed impact statement
would not be required if the assessment concludes
that the proposed action would have no significant
environmental impact. (354)
The Commission's directive stated that pending
completion of and public comment on the staff s
assessment of EPICOR-II, the licensee would not
be permitted to operate EPICOR-II except for
testing. (355) The statement added, however, that
the NEC's Director of NRR still could authorize
measures he deemed "necessary" to deal with an
"emergency" and that if he believed that public
health and safety required the use of EPICOR-II
before completion of the assessment, he would re-
port that to the Commissioners, who might permit
its use. (356)
This directive postponed operation of EPICOR-
II. The City of Lancaster and the NRC then
settled some of their differences. As spelled out in
a court order, filed May 29, 1979, the City agreed to
hold its pending motion for a preliminary injunc-
tion in abeyance. (357) The NRC was to prepare
an environmental assessment in accordance with
both the Commissioners' statement of May 25, 1979
and "such further terms and conditions as may be
provided by this court or further stipulation by
the parties." (358) Thus the order made the Com-
mission's May 25 directive a judicial directive as
well.
About the same time, four Pennsylvania resi-
dents and the Susquehanna Valley Alliance, an
unincorporated association of citizens whose
stated purpose is to preserve and protect the en-
vironmental quality of the Susquehanna River and
its environs, commenced a second injunctive action
in the same court. (359) They named the NRC,
GPU, Met Ed and a number of other parties as
defendants. (360)
The plaintiffs also sought an injunction against
the treatment of radioactive wastewater and its
discharge into the Susquehanna River. Like the
City of Lancaster, the plaintiffs held that the
NEPA required the NRC to prepare an environ-
mental impact statement or to make a specific
declaration that one was unnecessary. (361)
Among their other claims, the plaintiffs charged
there had been violations of the Federal Water
Pollution Control Act, 33 U.S.C. 1311 (f ) , which
prohibits the discharge of "high-level radio-
active waste into the navigable waters," (362) and
that the plaintiffs' constitutional rights had been
violated. (363)
In this case, the parties did not consent to a judi-
cial order based on the Commission's May 25, 1979
directive. The court did not grant the plaintiffs
injunctive relief.
On August 14, 1979, the NRC made public the
staff's environmental assessment of EPICOR-
202
II. (364) It covered only the environmental effect
of using EPICOR-II for processing radioactive
wastewater, concluding that the use of EPICOR-
II for this limited purpose would not "significantly
affect the quality of the environment." (365)
The staff deferred the more sensitive issue
how to dispose of the wastewater processed by
EPICOR-II, asserting that use of the system
for processing would not foreclose any options re-
garding ultimate disposal. (366)
The NRC sought formal public comment on the
assessment. (367) Among those who responded
were the city of Lancaster and the Susquehanna
Valley Alliance. Each contended that a detailed
environmental impact statement was necessary,
the assessment of EPICOR-II was not enough.
(368) Among its arguments, the Susquehanna
Valley Alliance said that the NRC was dividing
cleanup into eight segments, one of which was the
processing of wastewater through EPICOR-II,
(369) and that
[T]his segmentation is intended to create
the illusion that no single segment has any
potential significant environmental im-
pact, thereby negating the requirement of
preparing a full environmental impact
statement (EIS) covering the entire
[cleanup] program before the program
commences. (370)
The City of Lancaster also raised the issue of
improper segmentation. (371) In addition, the
City complained that the NRC assessment was an
. . . after-the-fact rationalization of the
particular decontamination alternative
which was chosen and constructed
prior to the preparation of the assess-
ment. (372)
The City urged that an environmental impact
statement be prepared by "an agency or firm not
associated with the nuclear industry or the NRC
staff." (373)
Not all the comments were negative. The Gov-
ernor of Pennsylvania forwarded an evaluation
performed by the State's Department of Envi-
ronmental Resources. It concluded that with some
specific exceptions, "the environmental assess-
ment is adequate and . . . EPICOR-II should
be used as soon as reasonably possible." (374)
Given the public's comments, the NRC staff
did alter some parts of the assessment; but it
did not change its conclusion that the operation of
EPICOR-II "will not significantly affect the
quality of the human environment." (375)
After getting the revised assessment, the Com-
mission received additional written comments, this
time from the Council on Environmental Qual-
ity (CEQ). (376) A statutorily created office
within the Executive Office of the President, the
CEQ is responsible for reviewing and apprais-
ing Federal environmental policies. It also pre-
pared the Federal regulations that spell out the
procedures for implementing XEPA. (377)
The CEQ observed that the XRC staff had pre-
pared one assessment for the processing of waste-
water, was going to prepare another on its dis-
posal, planned yet another on the release of radio-
active <rases from the containment and still had
other waste management issues to confront. The
CEQ expressed its concern "that the XRC staff's
review at TMI. as it is now planned, will result in
an inappropriate segmentation of the issues,"
It advised :
... it appears . . . several of the alterna-
tive operations being considered for TMI
Unit 2 will have significant impacts on
the environment. In these circumstances,
an environmental impact statement . . .
should be prepared. . . . (379)
Following this, the XRC and the CEQ held
meetings and then exchanged letters. One XRC
letter, dated October 15, 1979. warned the Council
that because of the continuing accumulation of
wa^tewater in the auxiliary building, "there is a
pressing need for action to deal with the inter-
mediate-level waste water." (380) The XRC noted
the two alternatives to EPICOR-II for decon-
tamination, but stated that each
... in effect would enlarge rather than
reduce the spread of radioactive contami-
nation and would involve potentially sig-
nificant safety questions and environ-
mental impacts. (381)
The letter also noted that the Commission had
concluded that "prompt decontamination of the
intermediate-level water by EPICOR-II is the
best response to the situation." (382)
The CEO responded to the XRC with a letter,
dated October 16. 1979. It clarified its position
concerning EPICOR-II :
Based on the assurances made in your
letter, the Council agrees that the prompt
decontamination of the intermediate-
level wastewater through the EPICOR-
II system is an operation necessary to
control the immediate impacts of an emer-
gency situation (40 C.F.R. 1406.11).
Xothing in this letter should, of course,
be taken as passing on the appropriateness
of other Commission actions thus far
under XEPA." 9 (383)
In its letter, the NRC had never specifically
called the situation an "emergency." Yet, in testi-
mony before the Subcommittee, CEQ's General
Counsel said that the Council had been "con-
vinced" an emergency existed, (384) warranting
the immediate action.
Xonetheless, the CEQ still "pressed" for prepa-
ration of a comprehensive environmental impact
statement covering all cleanup activities. (385)
The Commission issued a formal Memorandum
and Order, dated October 16, 1979, directing
prompt processing of intermediate-level waste-
water from TMI-2 using EPICOR-II. (386) It
also directed that the licensee maintain "suitable
tankage*' at Unit 1 that "could be used to store
wastewater from TMI-2 at an appropriate state
of readiness, should additional storage capacity
become necessary." 120 (388) In the Order, the
Commission maintained that, despite arguments to
the contrary, consideration of the impact of
EPICOR-II separate and apart from the overall
impact of the complete decontamination program
was not an "illegal segmentation." (389)
On October 12, 1979. the Federal district court
had dismissed the lawsuit of the Susquehanna Val-
ley Alliance on the ground that the plaintiffs'
claims first had to be presented to the NRC for
administrative review and determination before
the allegations could be considered "ripe" for anv
form of judicial review. (390) When the NRC f s
October 16 directive was issued, the Alliance im-
mediately appealed the dismissal to the U.S.
Court of Appeals for the Third Circuit, asking
that it review the district court's dismissal and.
pending this review, issue a judicial order pro-
hibiting operation of EPICOR-H. (391) Accord-
ing to an XRC attorney, the XRC opposed this
request for injunctive relief by arguing, in part,
that no water would be discharged into the river
and. as a result, the plaintiffs would not be harmed
by the use of EPICOR-IL (392)
The appellate court refused to halt the opera-
tion of EPICOR-II, although it did retain juris-
diction over the plaintiffs' appeal from the district
court's decision to dismiss the entire lawsuit. (393)
""The Council's narrowly drawn approval was based on the following Federal Regulation (40 C.F.R. section
1406.11) :
Where emergency circumstances make it necessary to take an action with significant environmental im-
pact without observing the provisions of those regulations, the Federal agency . . . should consult with the
Council . . . Agencies and the Council will limit such arrangements to actions necessary to control the im-
mediate impacts of the emergency.
""The Commission's Order did not explain why this particular requirement was included. The XRC's General
Counsel. Leonard Bickwit. and Commissioner Victor Gilinsky advised Subcommittee staff that the purpose was to en-
sure that additional tankage would be available for immediate use if EPICOR-II did not function properly and the
existing tankage in Unit 2 became filled to capacity with radioactive wastewater. (387)
203
EPICOE-II was finally used to process the
wastewater, some 5 months after it was ready.
At that time, the auxiliary building tanks were
some three weeks from capacity. 121
THE IMPACT STATEMENT
The NEC still had to decide what type of en-
vironmental studies to prepare on decontamina-
tion and waste disposal overall. On November 21,
1979, the Commission resolved the matter by issu-
ing a Statement of Policy and Notice of Intent to
Prepare a Programmatic Environmental Impact
tatement.
It announced that the agency would prepare a
programmatic environmental impact statement on
the decontamination and disposal of radioactive
wastes, observing that an "overall study . . . will
assist the Commission in carrying out its regula-
tory responsibilities ... to protect the public
health and safety as decontamination progresses."
(394) It noted that while the programmatic im-
pact statement was being prepared, the agency
was prepared to take prompt action, if needed:
For example, should the Commission be-
fore completing its programmatic state-
ment decide that it is in the best interest
of the public health and safety to decon-
taminate the high-level wastewater now
in the containment building, or to purge
that building of its radioactive gases, the
Commmission will consider ... [the
Council on Environmental Quality's]
advice as to the Commission's NEPA re-
sponsibilities. . . . (395)
The policy statement commented that "any action
of this kind" would not be taken without an "en-
vironmental review" and an "opportunity for
public comment." The statement also said :
. . . there may be emergency situations,
not now foreseen, which . . . would require
rapid action. To the extent practicable
the Commission will consult with [the
Council] in these situations as well. (396)
The NEC contracted with Argonne National
Laboratory to prepare the statement, at an esti-
mated cost of $2.5 million. As of early March 1980,
about 50 people were assigned to the project. In
late May, the draft statement was expected in
June, and the final statement was targeted for re-
lease between September and October 1980. (397)
As noted, in February an NEC Task Force in-
dicated the staff was not "clear" how the Commis-
sion intended to use the impact statement. As of
late Mav, the Commission still had to determine
how and by whom major cleanup decisions would
121 See "Technical Aspects of Recovery," p. 181.
be made after completion of the statement and the
statement's expected role in decisionmaking. (398)
The programmatic environmental impact state-
ment has created a dilemma.
On November 9, 1979, during Subcommittee
hearings, the NEC's then-Chairman, Joseph
Hendne, had predicted that the venting of kryp-
ton, like the decontamination of auxiliary building
wastewater, might become caught up in issues
of what could or should be done before completion
of an environmental impact statement.
The Chairman of the Subcommittee, Senator
Hart, had asked :
Short of an emergency, what do ... you
contemplate will happen to deal with con-
tainment water and trapped waste ? What
is an emergency and isn't? How much is
going to be helped from an EIS [environ-
mental impact statement] and how much
is not going to be in terms of cleaning this
operation in the next 6 months to a year?
(399)
Hendrie replied :
I think the place that we are going to have
a pinch is in dealing with the atmosphere
of the containment building as a necessary
preliminary step to getting on to process-
ing the water in the base of the contain-
ment, or a step that has to go along with
the processing of the water in the contain-
ment. Now what I would like to do is to
avoid the need for emergency action in
the sense that we just stop the environ-
mental review processes and say never
mind, we have got to do something, and
this is as cood a thing as we can see to do ;
so we do it. We went, in effect, through
that with the CEQ [Council on Environ-
mental Quality] on EPICOE-II because
things had just dragged on and there was
argument about whether we- there are
always people who want you to do five
more analyses. (400)
The issue arose again at a Commission meeting
on March 5, 1980, when Hendrie expressed his
views more strongly :
It is inconceivable to me that the laws of
the United States require us to sit on our
. . . [duffs] and fiddle for iy 2 years
waiting for that containment to leak
or the primary system to finally funk out
and fail to cool the core or the boron con-
centration to go. Don't we get . . . [re-
criticality] ? There has to be a way to get
in there and see that system is going to run
204
adequately for the balance of the time that
is necessary to clean up all the water, and
so on .
You can't .sit around here and calculate
environmental impact while we get ready
to have a disaster in central Pennsylvania.
I appeal to the staff, applicant, and God
for Christ's sake to tell me how to get
out of this idiocy.
Are we. in fact, compelled inextricably
under the laws of the United States to sit
here and wait for trouble ? (401)
Although preparation of an impact statement
requires extensive time and effort, it may help the
agency preparing it to focus on the alternatives,
and its procedures allow the public an opportu-
nity to comment as the document is being pre-
pared. (4O2) An impact statement provides one
means of deciding among competing interests
based on careful assessment of all alternatives.
Further, it is clear that the XRC can act in an
"emergency." although it is less clear under what
lesser circumstances it can do so, prior to comple-
tion of the statement. As the Commission, in re-
sponse to questions from the Subcommittee, stated :
... an overall environmental study of
the decontamination and disposal proc-
esses will not only assist the Commission
in discharging its regulatory responsi-
bilities to protect the public health and
safety but also assure that the public is
informed and. indeed, involved in the
Commission's decisionmaking process.
(403)
The Commission also noted that
... it is believed that such a statement
can serve as a useful planning tooL (404)
At another point in the Commission's March 5
meeting. Commissioner Hendrie elaborated on his
concerns:
This is the 1st of March and we are talk-
ing about the end of the year, that a
final EIS can be out and people begin to
complain about it and we will have to
fight court actions. It is not today, you
know, on the 5th of March. It is going to
be damn near a year from now and we are
still going to be sitting here . . . [star-
ing] at that containment. [H]ow many
neutron monitors do we still have on that
system \
' XRC STAFF: One.
HEXDRIF. : Anybody want to guarantee
me that it will still be there a year from
now \ Anybody want to guarantee me we
will know for sure what the vessel
boron concentration is based on the low
flows and taking the customary boron
sampling outside the building \ Anybody
going to be able to guarantee me we won't
have recriticality from low boron ... in
the next year \ How about breakdown of
the system inside ? (405)
Commissioner Gilinsky also expressed some
concern:
There ought to be, it seems to me, I think
a statement that deals with alternatives,
but it may be that we have gotten our-
selves into a very . . . elaborate state-
ment and certainly, the price tag seems
to suggest that. (406)
However, William Dircks, Acting Executive
Director for Operations, made the point that
". . . the impact statement, if it serves as a docu-
ment to help you plan action and carry out ac-
tions, ... is very important." (407)
Stephen F. Eilperin, Office of General Counsel,
explained to the Commission that the environmen-
tal impact statement need not delay cleanup :
The Commission's policy statement does
not [have to] await . . . the completion of
the programmatic statement to get into
the unit. (408)
In a meeting of the XRC on Xovember 29, 1979.
X'RC staff discussed the time required for an envi-
ronmental assessment or an environmental impact
statement. Yollmer commented :
The environmental assessment case would
add five months, if it is presumed that
one could allow venting of the contain-
ment as a method of cleanup. (409)
He went on to say
... if nothing could be done for cleanup
until the Environmental Impact State-
ment process is complete, then a mini-
mum of nine months would have to pass
before anything could happen. (410)
According to Vollmer, the environmental impact
statement would
. . . include everything that we can fore-
see, including fuel removal, waste dis-
posal everything we can see at this time.
(411)
REMOVAL OF THE KRYPTON GAS
On Xovember 13, 1979. Met Ed made a formal
presentation to the XRC on another major de-
contamination issue: how to remove the krypton
gas from the containment. The company asked
permission to purge the containment of the gas
over time, insisting that the "operation . . . can be
done with no significant hazard or radiation ex-
205
posure either to the general population or the
site." (412) Met Ed argued that
The time [required] to implement
[other] alternatives to purge are such
that we cannot guarantee full contain-
ment integrity and would, in fact, expect
general population doses to exceed those
minimum levels resulting from purge.
(413)
The NRC staff moved deliberately on Met Ed's
request. Denton noted :
I thought we had made great technologi-
cal strides when we found that we were
able to get the releases from this plant
following the accident within those of es-
tablished normal operating plants. Then
we were being sued by several communi-
ties not to permit releases that would
otherwise 'be acceptable within if the
plant had never had an accident.
So, we decided as a matter of policy to
look further to see if there was technol-
ogy available which would further re-
duce the impact of releases on the envi-
ronment . . . [W]e wanted to delay the
release of the krypton from the contain-
ment or water from the plant until alter-
natives could be explored and environ-
mental assessments could be prepared to
really be sure that we have looked hard
at the technology that might further re-
duce whatever the public impact would
be of release of this gas. (414)
As with EPICOR-II, the NRC staff did an en-
vironmental assessment on the question of the
venting. (415) It presented the assessment at a
Commission meeting on March 12, 1980. together
with its recommendation that controlled purging
was the preferred ootion. (416) Based on its as-
sessment, the staff concluded that purging
"would have no significant adverse impact on pub-
lic health and safety and no significant environ-
mental impact." (4i7) It also found
. . . that it is in the best interest of the
public health and safety to purge the re-
actor building promptly prior to comple-
tion of the Programmatic Environmental
Impact Statement. (418)
The final recommendation read :
We recommend that controlled purging
of the TMI-2 reactor building be author-
ized and that the licensee be directed to
propose a method for purging over a
shorter time period than the 60 days cur-
rently proposed, but within the con-
straints of Appendix I to 10 CFR 50
and 10 CFR 20. 122 (419)
The NRC staff suggested that the staff of the
Council on Environmental Quality might take a
different view of the NRC's authority to proceed
promptly with venting :
Until . . . [the environmental impact
statement] is prepared, CEQ staff be-
lieves that NRC approval of certain ac-
tions, such as purging the radioactive gas
from the containment, woulcl be a seg-
mentation of the entire clean-up program
in a manner inconsistent with . . . [the
National Environmental Policy Act]. 123
(421)
The NRC held a period of public comment on
the staff's recommendation for venting. It met
with substantial local opposition; some residents
expressed their complete lack of trust in the NRC
staff's recommendation. 124
As a result, at the end of March, Pennsylvania
Governor Thornburgh asked the Union of Con-
cerned Scientists (UCS) to study the proposal to
vent the krypton (422) and on April 11 sent a
letter to the NRC asking that the agency's period
for public comment be extended
... to reflect whatever facts or opinions
might emerge from this effort, and ac-
cordingly defer any final decision on the
cleanup proposal. 125 (424)
The NRC extended the comment period as re-
quested. (425)
On May 14, the Union of Concerned Scientists
released its results. The group concluded that the
venting proposal would not have any significant
adverse health effects but recommended against
the proposed venting method because of the stress
it would cause area residents. (426) It recom-
mended instead venting with the aid of a buoyant
plume or an extended vent stack using a plastic
tube supported by a balloon. (427)
la Appendix I to 10 C.F.R. Part 50 sets the guides and conditions by which the criterion "As low as is reasonably
achievable" is to be met in terms of radiation dose standards. 10 C.F.R. Part 20 defines the standards for protection
against radiation.
123 The NRC staff also noted, however, that the Council staff recognized that NEPA permitted the NRC to approve
certain actions that could result in "limited radioactive effluents" before completion of the impact statement. These ac-
tions included data-gathering activities and "actions necessary to maintain TMI in a safe and stable condition." (420)
124 See "Social Issues in Recovery," pp. 199-200.
"* In February 1980, the Pennsylvania Governor's Commission on Three Mile Island had urged the NRC to "make a
prompt decision concerning the proposed venting," adding that the Commission "would not oppose an NRC decision to
vent the krypton gas, provided that [projected] dose levels . . . are acceptable." [emphasis omitted] (423)
206
Two davs later, on May 16, Governor Thorn-
burgh sent a letter to Chairman Ahearne. citing
assessments from eight different sources, includ-
ing UCS. The Governor said :
There is, I have found, a broad-based
consensus among these sources that the
venting proposal now before you would
have . . . "no direct radiation-induced
health effects on the residents of this
area,"
* * *
Should you proceed with the venting pro-
posal advanced by your staff, be assured
that I am prepared to support that deci-
sion. (428)
In late May. the XRC staff again recommended
venting, finding that it was in the best interest of
public health and safety, would not have a signifi-
cant environmental impact and would not limit the
choice of reasonable alternatives for future cleanup
steps. (429) The CEQ concluded that as a matter
of procedure, based on those findings, the XRC
staff's proposal would not violate 40 C.F.R.
j 1506.1. which sets forth limitations on actions
during the XEPA process. (430) In early June,
the Xuclear Regulatory Commission formally
authorized venting. (431)
THE SITUATION, JUNE 1980
On March 17. 1980, the Third Circuit Court of
Appeals reversed the District Court's dismissal of
the Susquehanna Valley Alliance lawsuit. It con-
cluded that the lower court did have jurisdiction
to hear the plaintiffs' claims under the Xational
Environmental Policy Act and the Federal Water
Pollution Control Act, as well as plaintiffs' con-
stitutional claims. (432) So. in early June the
possibility of injunctive relief with respect to the
challenged decontamination activities was still
open. The appellate court's decision also set a legal
precedent for other parties who might want to
challenge cleanup proposals and activities in Fed-
eral court, including proposals for venting.
The City of Lancaster lawsuit was settled in
February. The XRC agreed not to allow the dis-
charge of water into the Susquehanna River be-
fore the end of 1981 without first complying with
the XRC's Xovember 21, 1979 policy statement.
The settlement spelled out the plaintiffs' right to
seek judicial review of any XRC decision to permit
discharges, including those authorized in the event
the XRC determined that an "emergencv" existed.
(433)
Work was continuing on the comprehensive
impact statement, a major undertaking that the
Commission had only decided to proceed with in
late Xovember 1979.' As of early June 1980, it
remained possible that most major decisions on
decontamination and waste disposal would be de-
ferred until the statement was complete. The final
statement was not expected to be completed and
ready to release until some time in September or
October 1980. (434)
The absence of decisions on firm plans for
cleanup has caused increasing concern. In late
January 1980, the Secretary of Pennsylvania's
Department of Environmental Resources, as
quoted in a newspaper article, announced that
TMI was "on its way to becoming one of the most
dangerous radioactive waste storage sites in the
world." (435)
Referring to the releases on February 11, 12 and
13, the Director of Pennsylvania's Department of
Environmental Resources, Bureau of Radiation
Protection, was quoted as saying:
We're going to see more and more of
this happening if we don't get in there
and clean that mess up ... We still have
an emergency situation at Three Mile
Island, and the XRC is treating it as if it
were a normal situation. It can't go on
like this for long before something gives.
(436)
CHANGES IN THE UNIT 2 LICENSE
In normal circumstances, maintenance and op-
eration of Unit 2 were governed by the facility's
Operating License and by the more elaborate con-
ditions set forth in the Technical Specifications
for Unit 2, a multi-volume document that detailed
the requirements of the license. Following the
accident, normal circumstances no longer existed.
Thus the XRC and the licensee recognized the need
to develop revised operating and contingency pro-
cedures to assure long-term cooling of the core and
plant stability. (437)
On July 20, 1979, the Director of XRR issued a
written order formally suspending the existing
license. The order directed that, pending further
amendment of the license, the licensee "maintain
the facility in a shutdown condition in accordance
with the approved operating and contingency pro-
cedures." It stated that XRC staff was preparing
"a detailed evaluation" of the license modifications
needed to "assure the continued maintenance of the
current . . . cooling condition," modifications which
would be set forth in new or revised Technical
Specifications. The JTRC anticipated having the
specifications available in a month. (438)
On January 11. 1980. after several deadline post-
ponements, the XRR sent the Commissioners an
order and revised Technical Specifications, to-
gether with an environmental assessment, which
concluded that the environmental impact of the
proposals would be "insignificant." (439) The
207
Comissioners gave their approval to NRR's sub-
missions and on February 11, 1980, the NRR
finally issued its order and revised Technical
Specifications. (440)
The order provided for the definition of operat-
ing parameters for long-term cooling and for im-
position of functional, operability, redundancy
and surveillance requirements concerning struc-
tures, systems, equipment and components needed
to maintain shutdown. (441) It also incorporated
the substance of the Commission's November 21,
1979 policy statement by directing that the license
be modified to :
[P]rohibit venting or purging or other
treatment of the reactor building atmos-
phere, discharge of water decontaminated
by the EPICOR-II system, and the treat-
ment and disposal of high-level radioac-
tivity contaminated water in the reactor
building, until each of these activities has
been approved by the NRC, consistent
with the Commission's Statement of
Policy and Notice of Intent to Prepare
a Programmatic Environmental Impact
Statement. (442)
The NRR order set forth procedures by which
the licensee "or any person whose interest may be
affected" could request a hearing with respect to
two issues : (1) whether the new requirements were
"necessary and sufficient for the maintenance of
the facility to protect health and safety or to mini-
mize danger to life and property" and (2) whether
the order "would significantly affect the quality
of the human environment." (443)
Requests for hearings were filed by two indi-
viduals and one organization, the Environmental
Coalition for Nuclear Power. (444) ENCP noted
that in April and May 1979 it had asked the NRC
to hold hearings on the "very issue of changes in
the Technical Specifications pertaining to the shift
from Operational Mode to Recovery Mode." (445)
IN SUMMARY
Once again, a review of events this time of the
regulatory proceedings reveals numerous di-
lemmas and unprecedented problems. The licensee,
for obvious reasons, wants to complete the cleanup
as quickly as possible, and in fact, the condition
of the plant suggests a need for prompt action. Yet
legal and regulatory procedures call for decisions
1M The proposed classes included : Class I all individuals, partnerships, corporations, institutions and other busi-
ness and professional entities within a 25-mile radius of TMI that suffered economic harm as a result of the accident ;
Class II all real property owners and residents within a 25-mile radius who suffered economic harm as a result of the
accident; and Class III all individuals within a 25-mile radius who suffered personal injury, incurred medical ex-
penses, suffered emotional distress or will require medical services to monitor the possibility of latent defects from
exposure to radiation. (450)
The damages allegedly sustained include "a substantially increased probability of incurring cancer and/or genetic
defects because of exposure to radiation; damages associated with the necessity of evacuation; reduction in the finan-
cial value of property and business; contamination or spoilage of products; and work stoppages." (451)
to be made deliberately after weighing alterna-
tives and affording opportunities for public com-
ment. The NRC has followed the necessarily de-
liberate procedures established to achieve these ob-
jectives, while maintaining the right to act
promptly under certain conditions. As it proceeds
in this fashion, the NRC must deal, as must the
utility, with the distrust and vocal opposition of
many residents around Three Mile Island.
FINAL COST TO THE LICENSEE
Judicial and regulatory proceedings will also in-
fluence the cost of the accident to the licensee.
JUDICIAL PROCEEDINGS
Civil Tort Actions
Following the accident, Met Ed, Jersey Central,
PENELEC and GPU were named as defendants
in numerous civil lawsuits brought by private
parties seeking to recover for alleged personal and
property damage. (446) Plaintiffs have charged,
among other things, negligence or willful miscon-
duct with respect to the design, construction,
operation and maintenance of the TMI facility and
with respect to the nature of the information re-
leased to the public as events progressed. (447)
Strict liability claims also have been asserted,
based on the alleged "miscarriage of an ultra-
hazardous activity," namely the operation of a
nuclear reactor. (448)
Early in the proceedings, for reasons of judicial
economy, most of the lawsuits were consolidated
in Federal District Court into a single class ac-
tion lawsuit called Fantasky v. GPU. (449) More
than 60 named plaintiffs, representing businesses,
property owners and residents, are part of
Fantasky. The plaintiffs proposed to represent
not only themselves, but similarly aggrieved
parties who might not bring their own lawsuits. 126
This class action suit requests monetary dam-
ages, an order "directing that the [TMI] nuisance
be abated," and imposition of a "constructive
trust" on property owned by the defendants to pay
for the cost of medical diagnosis and treatment of
"possible cancerous and abnormal genetic con-
ditions." (452)
Early in the proceedings the parties in Fan-
agreed that the plaintiffs may represent
208
Price-Anderson, however, will not necessarily
limit the total exposure of the GPU companies to
$15 million. According to GPU's 1979 annual re-
port, the lawsuits for personal and property dam-
ages (including claims for punitive damages) and
the injunction actions have raised questions
whether certain claims "material in amount" are
subject to the liability limits of Price- Anderson or
are outside the insurance coverage provided pur-
suant to the statutory scheme. (463)
Price- Anderson has other provisions that may
be pertinent to the pending civil tort actions. In a
severe nuclear incident, described as an "extraordi-
nary nuclear occurrence," Price-Anderson pro-
vides that a licensee covered by the statute's system
of financial protection may be required to waive
certain legal defenses. (464) When the waiver
occurs, the plaintiff no longer has to prove the
licensee's negligence and, in addition, may insti-
tute his action at any time within three years from
when he "knew, or reasonably could have known"
of his accident-related injury or damage, so long
as the action is not begun more than 20 years after
the incident. (465) The plaintiff still must prove
injury or damage, the monetary amount of the loss
and the causal link between that loss and the nu-
clear accident.
Price- Anderson assigns the XRC responsibility
for making the critical judgment whether an ac-
cident is an "extraordinary nuclear occurrence" or
"EXO." The statute defines an EXO as
. . . any event causing a discharge or
dispersal of source, special nuclear, or
byproduct material from its intended
place of confinement in amounts offsite
or causing radiation levels offsite which
the Commission determines to be sub-
stantial, and which the Commission de-
termines has resulted or will probably re-
sult in substantial damages to persons
offsite or property offsite. (466)
and adds that the
Commission shall establish criteria in
writing setting forth the basis upon
which the determination shall be made.
(467)
The Commission's criteria are embodied in the
Federal Regulations at 10 C.F.R. Part 140.
On July 23. 1979. the XRC published a notice
that it was initiating proceedings to make an
EXO determination. (468) On August 17 the
is presently set at $560 million according to the formula set forth in the statute at 42 U.S.C. section
) Price-Anderson also provides that Congress may take "necessary and appropriate" action-such as pro-
ding for -additional payment to claimants if damages exceed $560 million. 42 U.SC section 2210e
If that l" PxhVi * 14 miUi ^ n in artli tJ-Purchased private insurance is available for the accident,
that is exhausted, the utilities may be assessed a maximum of $335 million. If that, too, is exhausted the Govern-
roeram covers a maximum of $85 million. With respect to the $335 million from utilities, the amount
nuclear reactor plant licensees so that each pays a premium of $5 million. (461)
it i involved in the accident thus is assessed no more than any other licensee. Since the GPU
three operating licenses, their maximum assessment would be Slo million.
all aggrieved parties falling within Classes I and
II. (453) The plaintiffs' attempt to represent Class
III members was disputed. The issue, together
with a U.S. magistrate's recommendation, has been
submitted to the District Judge, but as of late May,
no decision had been rendered. (-454)
In addition to the Fantasky consolidated class
action, many other civil tort actions still were
pending in late May 1980. There was. for example,
a second class action involving dozens of plaintiffs
seeking both monetary damages of at least $560
million and also punitive damages. (455) More-
over, another lawsuit was pending involving a
couple who alleged that radioactive releases from
TMI during the accident had caused the stillbirth
of their daughter some 5 months later. (456)
For reasons of judicial economy, all suits filed
since the Fantasky class action have been consoli-
dated with Fanta$ky. absent a showing that a suit
should be treated separately. (457) This will per-
mit consideration of common legal and factual
issues in a single proceeding.
It is difficult to predict how long the civil tort
will last. In late Mav. more than one year
after the accident, the Fantasky consolidated
class action was still in a relatively early, proce-
dural stage. (458) Because of the dispute over
Class III plaintiffs, notice to the prospective class
had not yet been provided. Xo final adjudication of
the class action litigation is possible until such
notice is given so that prospective class members
have an opportunity to advise the court whether
they wish to be represented by the plaintiffs. Liti-
gation could continue for years after the notice is
d.
Although the civil tort actions involve substan-
tial sums of money, the potential financial burden
imposed on the GPU companies is limited in part
by the Price- Anderson Act. as amended. (459)
which is designed in part to reduce the financial
exposure of any one licensee in the event of a
nuclear accident. Price- Anderson limits the total
amount of claims that must be paid to persons in-
jured in a "nuclear incident" 12T and provides for
these claims to be covered under a system of utility-
purchased private insurance, retrospective premi-
ums assessed against the utilities, and government
indemnity. (400) Under this system of financial
protection. Met Ed. Jersey Central and PEXE-
"2 do not expect to be assessed more than $15
million in retrospective premiums for all the TMI-
generated public liability claims. 128 (462)
209
Commission formed a panel of staff to assemble
the relevant information, evaluate public com-
ments and report to the Commission its findings
and recommendation. (469)
In December 1979, the staff sent its report to
the Commission. The panel concluded :
. . . that the first criterion, pertaining to
whether the accident caused a discharge
of radioactive material or levels of ra-
diation offsite as defined in 10 C.F.R.
140.84, has not been met, [The panel]
. . . further finds that there is presently
insufficient information to support any
definitive finding as to whether or not the
second criterion, relating to damage to
persons or property offsite as defined in
10 C.F.R. 140.85, has been met. Since
the Panel has not found that both
criteria have been met, it recommends
that the Commission determine that the
accident at Three Mile Island did not
constitute an "extraordinary nuclear oc-
currence." (470)
In April 1980, the Commission made a final
determination that the accident did not constitute
an "extraordinary nuclear occurrence," as defined
by the Price-Anderson Act and the Commission's
regulations. Noting that "in ordinary parlance"
the accident was "extraordinary," the Commission
nonetheless found that the radiological releases
associated with the accident did not rise to the
levels required for an ENO determination. (471)
Even assuming the plaintiffs are not able to
prove negligence in their tort actions, the Commis-
sion's conclusion will not necessarily have a deci-
sive effect on the outcome of these lawsuits. In
addition to alleging negligence, the plaintiffs have
also been arguing that defendants are strictly liable
under State law, that is, without regard to whether
they acted negligently. (472) A finding of no ENO
will have no legal effect on these separate strict
liability claims. Moreover, GPU and its subsidi-
aries may not insist that the plaintiffs prove negli-
gence. Court papers filed well before the Commis-
sion's determination indicated that the defendants
in Fantasky had made an undertaking not to re-
quire any person claiming compensatory damages
for personal injury to prove negligence and might
be willing to make a similar agreement with plain-
tiffs alleging economic loss. (473)
Stockholder Suits
GPU is also a defendant in litigation instituted
by its own stockholders. Two class actions were
brought in Federal District Court against GPU
and a number of the companies' directors on be-
half of GPU stockholders. (474)
The lawsuits include charges that defendants
128 See "Prior to the Accident," pp. 77-78.
;
violated the securities laws by failing to disclose to
stockholders and the public defects in the design,
installation and operation of TMI Unit 2, all of
which allegedly were known by defendants prior
to the accident. As a result of the alleged non-
disclosures, plaintiffs said they purchased GPU
stocks at inflated prices. (475)
For judicial economy, these stockholder class
actions were consolidated in the U.S. District
Court for the District of New Jersey. (476) The
District Court certified a class that includes pur-
chasers of GPU common stock from August 25,
1975 through April 1, 1979. (477) As of mid-
March, notices had not gone out to the class.
GPU Suit Against Babcock & Wilcox
In one noteworthy instance, GPU instituted
its own lawsuit as a result of the accident. In
March 1980, GPU and its three utility subsidiaries
commenced a civil damage action in Federal Dis-
trict Court against Babcock & Wilcox (B&W),
the nuclear reactor supplier for TMI, and against
B&Ws parent company. (478) One newspaper
article described the suit as a "jarring break in
what had been a united industry front on nuclear
power questions." (479)
Asserting four separate causes of action, the
complaint charged ( 1 ) gross negligence and reck-
less disregard of foreseeable consequences or, in
the alternative, ordinary negligence; (2) strict
liability because of the risks and consequences of
an accident resulting from defects for which B&W
was responsible; (3) breach of contract; and (4)
breach of implied warranties. (480)
As respects their negligence claims, the plain-
tiffs alleged, in part, that B&W had received
"prior warnings" of problems as a result of "simi-
lar incidents" at the Davis-Besse plant. 129 They
also alleged "inadequacies" in the B&W nuclear
steam supply system, related equipment, limits and
precautions, procedures and training. (481)
As damages, plaintiffs cited, among other items,
the expense of purchasing replacement power,
cleanup costs and the loss of a reasonable return on
capital invested in Unit 2. Plaintiffs' complaint
said that damages had exceeded $500 million with
the anticipation of "very substantial future dam-
ages." (482)
NRC REGULATORY PROCEEDINGS
Civil Penalties
Seven months after the accident, the Director of
the Office of Inspection and Enforcement (I&E)
served Met Ed with a Notice of Violation and a
Notice of Proposed Issuance of Civil Penalties.
Based on its investigation of the accident, I&E
described a number of instances of "apparent non-
210
compliance" with NEC's regulations, the Techni-
cal Specifications for Unit 2 and the procedures
mandated by the Technical Specifications. (483)
I&E cited six "violations," ten ''infractions'' and
one "deficiency." Most of the proposed penalties
related to a single violation Met Ed's failure to
block off the pilot -operated relief valve on the re-
actor's pressurizer from October 1978 until some
two hours after the accident began on March 28,
1979. 130 (484) For this alleged violation, each day
of non-compliance was treated as a separate offense
subject to a $5,000 penalty, resulting in a cumula-
tive civil penalty of $630.000. (485)
The total amount of civil penalties for all items
added up to $725,000. (486) However, by statute,
the maximum assessable civil penalty for any 30-
day period was $25.000. 131 (487) Since the viola-
tions related to a five-month period from October
1978 through March 28. 1979, I&E's proposed pen-
alty thus was reduced to $155.000. (488) It was still
the largest civil penalty the XRC had proposed up
to that time.
On January 23. 1980. after receiving Met Ed]s
response to the items of "apparent non-compli-
ance" and proposed penalties (489). I&E issued a
formal order imposing $155,000 in civil penalties.
(490) On February 14, 1980. Met Ed paid the fines,
foregoing its right to a hearing. (491)
Imposition and payment of the $155.000 in civil
fines did not bring an end to the XRC's considera-
tion of accident-related penalties against Met Ed.
On March 4. 1980, the XRC's Special Inquiry
Group, which had previously investigated and re-
ported on the accident for the Commission, sub-
mitted a supplementary report to the Commission
Chairman concerning whether Met Ed officials had
intentionally withheld information from the XRC
as the accident unfolded on March 28, 1979. (492)
The supplementary report said that there was in-
direct evidence from which one could infer that
information was intentionally withheld, but that
the record, taken as a whole, did "not permit the
unbiased observer" to reach this conclusion "based
on actual evidence." (493) With this report in
hand, the XRC formed a group to assess the ade-
quacy of the Special Inquiry's work on this issue
and to determine whether further action, such as
civil penalties, might be required. (494) As of
May 1080. this assessment was still continuing.
XRC staff members also have considered the
charges of Harold Hartman. a former TMT con-
trol room operator, who alleged that for months
prior to the accident, Met Ed employees had been
falsifying test data on the rate of leakage of the
same pilot-operated relief valve that stuck open on
the day of the accident. 132 According to a report
published May 15, 1980, a separate Federal Grand
Jury investigation began into Hartman's charges.
The report suggested that while the Grand Jury
was questioning Met Ed employees as individuals,
the inquiry might be expanded to include Met Ed
management. Pending completion of the Grand
Jury investigation, the XRC will not pursue in-
formation being looked into by the Grand Jury.
(495)
In April 1980, civil penalties also were assessed
by the Director of I&E against Babcock & Wil-
cox (B&W). It was the first time such a penalty
had been proposed by XRC staff for a company s
activities as a reactor supplier. (496) The pro-
posed fine, which totaled $100.000. 133 was based on
four items of non-compliance. Each cited item re-
lated to B&Ws alleged failure to evaluate and
report on significant safety information, including
information set forth in the Michelson Report, 1 **
in violation of 10 C.F.R. Part 21. (497) In trans-
mitting the charges to B&W, I&E's Director
charged generally that B&W "did not have an
effective system for collection, review and evalua-
tion, and reporting of important safety informa-
tion." (498) In its May 20, 1980 response, B&W
denied the charges, but paid the fine, saying that
further proceedings would be "tune-consuming,
expensive and needlessly divert" the attention of
"critical personnel and resources." (499)
Suspension of TMI-2's License
Another matter pending before the NBC is the
status of the Unit 2 Operating License. On July ,20,
1979, the N3JR, as noted, had issued an order for-
mallv suspending this license. On October 25. 1979,
the Commission held a meeting at which the issue
of the license was discussed. Commissioner Gilin-
sky recommended that the Unit 2 license should be
revoked because revocation, unlike suspension,
would be "a very strong statement" of the Com-
mission's position. (500) Suspension, in his view,
was only "an intermediate step between not taking
action and revoking licenses." (501) Commissioner
Ahearne agreed that revocation would "be seen as
different by the public," (502) but argued that
"revocation of TMI-2's license is not meaningful.
"" See "Prior to the Accident." pp. 71-72. for a discussion of the leakage.
m Both the House and Senate, in action on the XRC authorization Mil for fiscal year 1980. passed provisions that
would allow XRC to impose civil penalties of up to $100.000 per violation and would eliminate any limitation on the
penalty amount assessable in a 30-day period. As of mid-May the authorization bill had been agreed to in conference
and was awaiting enactment.
m See 'Prior to the Accident." p. 71. fn. 49. for more details on the charges made by Hartman.
Each day of non-compliance was treated as a separate violation subject to a $5.000 penalty, resulting In a cnmo-
lative penalty of $575.000. As was true of the penalty against Met Ed, however, the $25,000 statutory limit for any 30-
day period reduced the assessable penalty to $100,000.
"* See "Prior to the Accident," p. 78, for a discussion of this report.
211
given the status of that system." (503) In
Ahearne's view :
I do not think it would be seen as differ-
ent in substance by any of the people who
are knowledgeable with the proceedings
or the fact of the plant or any of that side
of the nuclear industry. I think it will be
perceived by the industry side . . . even
. . . the public interest side who are famil-
iar with it as an attempt by the Commis-
sion to position itself in a way that makes
it look as though they're taking a strong
stance. (504)
The Commissioners voted on whether to revoke
Unit 2's license. Gilinsky and Bradford voted for
revocation, Hendrie and Ahearne against it j Com-
missioner Kennedy was not present. The tie vote
meant that the Unit 2 license would remain sus-
pended. (505)
Restart of Unit 1
The agency also has to decide what to do about
the operation of Unit 1. On March 28, that facility
was about to resume operation after being out of
service for refueling. It has remained shut down
since, initially because of attention to the accident,
then because it was subject to an NEC shutdown
order for plants with the B&W nuclear steam
supply systems used in Unit 2. The NEC subse-
quently permitted other affected plants to resume
operation. At that time, Met Ed advised the NRC
that it would not restart Unit 1 without provid-
ing advance notice. (506) Shortly thereafter, on
June 28, 1979, the licensee informed the NRC of
various actions it proposed to take prior to re-
starting Unit 1, including,
all those [actions] . . . proposed or re-
quired in respect of the other B&W units,
as well as additional actions that Met Ed
believed appropriate. (507)
On July 2, 1979, the Commission ordered the
facility to remain in cold shutdown until further
notice. (508) On August 9, the Commission issued
another order explaining its action. (509) Beyond
the questions relating to the B&W design, the
Commission identified several other issues requir-
ing resolution, including the potential interaction
between Unit 1 and the damaged Unit 2, Met Ed's
management capabilities and technical resources,
the potential effect of decontamination operations
on Unit 1, and the "recognized deficiencies" in the
licensee's emergency plans and operational proce-
dures. (510)
The Commission specified short- and long-term
actions needed to resolve some of the concerns
Unit 1 was to stay shut down pending "satisfac-
tory completion of the short-term actions and
reasonable progress" toward completion of the
long-term ones. (511) The Commission designated
the Atomic Safety and Licensing Board to conduct
hearings and render an initial determination on
the resumption of Unit 1 operations. The NRC in-
dicated that the Board's recommendation would be
transmitted directly to the Commission for its final
decision. (512)
In the following months, the Licensing Board
ruled on petitions from parties wanting to inter-
vene, determined what contentions would be
heard, and reviewed other pre-hearing matters.
Among those permitted to intervene were the
Commonwealth of Pennsylvania, the County of
Dauphin, Pennsylvania, the Pennsylvania Public
Utility Commission, the Union of Concerned
Scientists, and a number of other organizations
with members residing near TMI. (513)
When the Unit 1 restart hearings begin, the
Board will take evidence on a number of issues,
among them whether the licensee's decontamina-
tion and restoration work on Unit 2 can be com-
pleted without affecting the safe operation of Unit
L and whether the licensee's financial condition
might undermine its ability to operate Unit 1
safely. (514) As of late May, the Board had not
decided whether it would also consider the psycho-
logical distress of citizens living near TML 135
In its August 9, 1979 order, the Commission said
it expected the Board to conduct the proceeding
'expeditiously." It set initial "milestones," calling
for the Unit 1 restart hearings to begin about
February 1980. (517) Yet, as of late May 1980, it
was unlikely that hearings would begin before the
fall, 136 (521) and no firm date for a final decision
had been set.
OTHER REGULATORY PROCEEDINGS
Met Ed and PENELEC are regulated by the
Pennsylvania Public Utility Commission (PUC)
and Jersey Central by the New Jersey Board of
Public Utilities (New Jersey Utilities Board). As
discussed earlier, the regulatory commissions have
been determining how much customers must pay
for their power while Units 1 and 2 are out of
service.
On June 15, 1979, the Pennsylvania PUC re-
moved from Met Ed's and PENELEC's rate bases
all costs associated with Unit 2, including clean-
In late February 1980, the Licensing Board recommended to the Commission that evidence on the issue of
K g ?A^ eSS bC K a ? en . d Vl lng the restart hearin ^. (515) The Commonwealth of Pennsylvania, among others,
had ^fff. "mt the psychological health of residents had to be considered in deciding whether to restart Unit 1. (516)
the \ L Tnnn^i iof '/Jj e , llCe ? S f es " ta l get " date for restartin g Unit 1. assuming a' favorable decision from
i iL! <Ja / e 1 ( ; onslde 1 red optimistic by some. (519) Before the XRC decided to hold restart
had talked of restarting no later than January 1, 1980. (520)
212
up repair, disposal of wastes and decontamina-
tion. (522) It stated that a utility is entitled to
charge rates permitting a fair return on property
that is "used and useful in the public service" and
that Unit 2 was no longer "used and useful." (523)
It explained :
There is a great uncertainty with respect
to when, and in fact if ever, TMI-2 will
resume operation. Respondents estimate
that TMI-2 will be out of service for two
to four years. However, no one has been
able to determine the extent of damage to
the fuel core. Design and operation
changes may be ordered by the Nuclear
Regulatory Commission, but these are as
yet unknown. Public sentiment has been
expressed against the renewed operation
of TMI-2 : and the cost of repair, cleanup
and waste removal may be so high as to
make restoration of the plant uneco-
nomic. (524)
The Pennsylvania PUC did not reach the same
conclusion for Unit 1 at this time. Noting that
GPU's president had said Unit 1 could be generat-
ing power "as early as August 1979, and certainly
no later than January 1, 1980," the PUC con-
cluded that "TMI-1 is at present only experienc-
ing an outage" and would not be removed from
the rate base. 137 (526) However, the PUC said it
would "monitor the status'" of Unit 1, and if start-
up were delayed beyond January 1, 1980, it would
begin proceedings to decide whether Unit 1
should remain in the rate base. (527)
At the same time as it removed Unit 2 costs from
the rate bases, the PUC held in favor of the two
utilities on the important issue of replacement
power. Before the accident. Units 1 and 2 had
provided roughly 30 percent of the energy of the
GPU system. u.2S) After the accident, the utili-
ties continued to provide electric service to their
customers by purchasing power from other
sources. (529) Among them was the so-called
Pennsylvania-New Jersey-Maryland Interconnec-
tion. 13S a utility pooling arrangement that per-
mits bulk purchases at reduced rates. (531)
In its June 15 decision, the Pennsylvania PUC
granted rate relief to help meet replacement power
costs, 139 The PUC reasoned that if the utilities
had not bought replacement power, they would
have had to reduce service to consumers or increase
use of the utilities' existing plants, "many of
which have higher operating costs than the costs
of purchased power." (533) The PUC found the
power replacement purchases "to be in the public
interest," (534) and said that:
The purchase of energy is a reasonable
and necessary cost of providing service
which must be recovered from rate-
payers. Service cannot be provided with-
out cost. It is equitable for the ratepayers
of Met Ed and PEXELEC to pay the
costs of purchasing power since they are
receiving service and will be paying none
of the costs of TMI-2. (535)
The PUC further emphasized that :
[T]he total rates for electric service to the
customers of Met Ed and PEXELEC will
be no greater than the rates which would
have been allowed had the incident never
occurred. 140 (538)
By September 1979, it had become apparent that
Unit 1 would not be back in operation before
January 1, 1980. The Pennsylvania PUC there-
fore commenced a proceeding to determine
whether costs associated with that unit should be
removed from the rate bases of Met Ed and
PEXELEC. (539)
In their formal response, the utilities blamed
the NRC, claiming they had been trying to con-
vince the agency to adopt procedures permitting
an early restart of TMI-l. (540) They charged
"discriminatory action" :
Respondents have been, and are, totally
unable to understand how the XRC could
so disregard the national and public in-
terests involved in permitting restart of
TMI-l as early as it can be demonstrated
that such restart is consistent with the
public health and safety. (541)
The utilities noted that their existing rates were
neither the lowest in the Commonwealth, nor the
highest. (542) The utilities also raised the prob-
lem of cash flow. They stated that they had "had
to borrow substantial amounts*' from the banks to
provide the cash needed to purchase replacement
power, and that removal of Unit 1 costs from the
rate base might adversely affect the willingness of
^ The GPU president's prediction was made before the NRC directed that Unit 1 remain shut down. The State
order also predated the NRC's action agrainst Unit 1. i 525)
m The GPU companies also were able to make bulk purchases from other power supplies. In July through November
I'.t7',. fur example, they received substantial energy from outside the Interconnection pool. (530)
"The relief granted amounted to about 85 percent of actual replacement power costs. (532)
^According to the PUC. this conclusion was based on a comparison of average revenues from the rates set in its
Order with average revenues derived from base rates including the costs of Unit 2 and energy rates charged prior to the
accident. (536) In November 1979. GPU's president testified that the PUC's rate decisions meant that customers were
paying essentially what they would have paid had Unit 2 never been built. (537)
213
these lenders to continue to provide the necessary
cash: 141 (543)
Respondents will continue to take all
actions available to them to continue to
render adequate, reliable service to their
customers. . . . But Respondents do not
possess the ability to ensure that such
service will be rendered. The action taken
by ... [the Pennsylvania PUC], and the
response of the banks to that action, are
major determinants of both the adequacy
and the cost of such service. (544)
In early November 1979, the Pennsylvania PUC
commenced another proceeding, this time just
against Met Ed, to determine whether that utility
should lose its certificate of public convenience
its franchise to provide electric power in Penn-
sylvania. (545) The PUC noted that Met Ed was
likely to incur substantial expenses as a result of
the accident, and also that the President's Com-
mission had found "a number of important
cases" before the accident in which GPU and Met
Ed had been guilty of "a serious lack of com-
munication about several critical safety matters"
relating to the operation of Unit 2. (546) Accord-
ing to the PUC, there thus were :
. . . serious questions about the continued
ability of Met Ed to provide safe, ade-
quate, and reliable electric service at just
and reasonable rates. The Commission,
therefore, finds it in the public interest to
put at issue . . . the continued viability of
Met Ed as a public utility. (547)
The PUC consolidated the issue of Met Ed's
viability as a utility with the issue of TMI-l's
"used and useful" status and also with a request
made by Met Ed on November 1, 1979 for addi-
tional rate relief to cover increased replacement
power costs. (548)
In December, the Pennsylvania PUC began for-
mal hearings on the three issues.
On February 8, 1980, with hearings still contin-
uing, the Pennsylvania PUC granted Met Ed an
interim rate increase to meet its higher replace-
ment power costs. The increase provided Met Ed
an estimated $55 million during 1980, (549) but
was made subject to adjustments reflecting the
final results of the PUC's inquiry. (550)
The interim rate relief had followed a decision
by the PUC that its proceedings would not be
completed until May 23, 1980. In its interim rate
order, the PUC said :
[W]e do not intend to engage in 'brink-
manship. The present financial condition
of Met Ed is too serious a matter and of
too great importance to the public Met
Ed serves to warrant the risk of further
financial burden brought on by delay
arising from the inability of the parties
to meet the intended schedule of the Com-
mission. We are convinced that the public
interest requires that this Commission
provide Met Ed's bank creditors with the
requisite assurance that they can ulti-
mately be repaid. (551)
On May 9, 1980, after twenty-seven days of
hearings, the Pennsylvania PUC rendered its
initial decision on the three issues before it. 142
Describing these issues as "exceedingly difficult"
to resolve, the PUC said that it
. . . has had to balance the need to explore
and carefully examine Met Ed's contin-
uing, long-term viability against the ur-
gency to act promptly to avoid being
overtaken by events. In addition, the
Commission has had to resolve the com-
peting concerns of creditors who want
assurances of earnings and ratepayers
who want equity in allocating the costs
associated with the . . . accident; and who
see an inequitable duplication in paying
the costs of TMI-1 and the costs of
TMI-1 replacement power; and of ...
[the utilities] who would emphasize their
financial needs and other parties seeking
a determination based on other economic,
social, and political principles. (553)
The PUC's conclusion, it said, was that "Met
Ed should continue to operate as a public util-
ity/' (554) The PUC described its order as pro-
viding
... an adequate framework for Met
Ed's recovery. Respondent must convince
its bank creditors that it has the will and
the ability to rehabilitate itself. (555)
The PUC's decision criticized the Federal Gov-
ernment :
Regretably, the Commission must
again decry the failure of the Federal
Government to respond to the accident at
Three Mile Island with financial assist-
ance that is commensurate with its re-
sponsibility for nuclear energy . . . The
people of Pennsylvania should not have
to bear the entire burden emotionally
or financially where that burden prop-
erly belongs to all those who have bene-
141 See "Financial Aspects of Recovery," pp. 191-193, for further details about this lending arrangement and its fi-
nancial implications.
M2 This was an "initial decision." After a two-week period for the filing of exceptions by the parties, a final order
with changes not pertinent to this discussion, was issued on May 23, 1980. (552)
214
fitted from the development of nuclear
energy. *
. . . [W]hat is painfully clear is that an
economic catastrophe has befallen the
GPU Companies, and their ratepayers
and investors as well. We believe that
Congress has a parallel responsibility to
act in this situation, noting that when the
prospect of a nuclear "incident" seemed
remote, Federal willingness to render as-
sistance to the nuclear industry was free-
flowing. Now that such a tragedy has be-
come more than a remote possibility, that
willingness has dissipated. Never has it
been more true that victor)' has a thou-
sand followers, but that defeat is an or-
phan. (556)
Specifically, the PUC declined to revoke Met
Ed's Certificate of Public Convenience "because
we find no imminent and foreseeable threat to con-
tinued provision of adequate and reliable serv-
ice at reasonable rates." (557) However, the PUC
left open the possibility that it would consider the
issue again if necessary. (558)
Second, the PUC removed capital and operat-
ing costs associated with Unit 1 from the base
rates of Met Ed and PENELEC on the ground
that Unit 1 was no longer "used and useful" in
the public service. In explaining its decision, the
PUC noted the ongoing NRC proceedings re-
garding restart of Unit 1 and said that there was
"substantial uncertainty'' as to when or whether
the facility would be returned to service. 143 It
added, however, that if and when the NRC al-
lows the restart of Unit 1. the PUC would give
"priority treatment" to reconsidering its determi-
nation on this issue. (560) The PUC also said that
Met Ed must aggressively pursue the
return to sen-ice of TMI-1 or an early de-
cision on its conversion and use of an al-
ternative fuel. (561)
Third, the PUC concluded that Met Ed and
PEXELEC should have rate relief needed to per-
mit full and current recovery of their replacement
power costs. 144 The Commission said that its de-
termination to do so was "inseparably inter-
twined" with its decision to remove Unit 1 costs
from the utilities' rate bases; and that the rate
relief should lessen the utilities' need for short-
term borrowing and facilitate the utilities' efforts
to obtain permanent financing. (563)
Fourth, the PUC determined that the utilities
should receive additional rate relief to permit
them to recover over an eighteen month period cer-
tain energy costs that had not previously been
covered through rate-making, including previ-
ously unreimbursed replacement power costs.
(564)
New Jersey's Board of Public Utilities has been
similarly involved in rate-making issues. The
Board regulates Jersey Central, which, like PEX-
ELEC, shared in the cost of operating Units 1 and
2 and prior to the accident drew power from the
TMI facilities.
On June 18, 1979, the New Jersey Board of Pub-
lic Utilities took much the same approach as the
Pennsylvania PUC. (565) The Board concluded
that TMI-2 was not "used and useful" in providing
service to customers and reduced Jersey Central's
rate base by $29 million; it refused to take out
TMI-1 costs, finding that "the outage of this facil-
ity is of a temporary duration" ; and it permitted
Jersey Central to recover about 85 percent of its
estimated replacement power costs. (566) The
Board also ordered the utility not to pay any divi-
dends to its parent, GPU, for the remainder of
1979. (567)
Early in 1980, Jersey Central requested addi-
tional rate increases, some but not all of which re-
lated to TMI costs. The request led the New Jersey
Board to consider whether Unit 1 costs should be
removed from Jersey Central's rate base and
whether the utility should receive additional rate
increases for TMI-related replacement power
costs. (568)
On April 1, 1980, the Board rendered two related
decisions. In the first, it granted a rate increase for
an eleven-month period to cover replacement
power expenses associated with Unit 1. (569) In
the second, the Board removed costs associated
with Unit 1 from Jersey Central's rate base on
the ground that the facility was not "used and
useful." Like the Pennsylvania PUC, the Board
concluded that the NRC restart proceedings would
keep TMI-1 out of service for an extended period,
at least two years the Board estimated. The Board
added that "if and when" Unit 1 is returned to
service, the Board would "expeditiously return the
unit to ... [the] rate base." (570) Having removed
Unit 1 costs from the rate base, the Board approved
other action to help soften the financial effect. It
lu Although the utilities testified to an estimated in-service date of January 1, 1981, there was testimony from the
Pl'C's consultant that mid-1983 was a realistic start-up date for Unit 1. The PUC also noted that the NBC's restart
proceedings would be considering, in part, whether Unit 1 could he safely operated before completion of Unit 2 cleanup
and that a GPU official had estimated that cleanup would not likely be completed until after June 1983. (559)
1M The PUC added that not "every dollar of purchased power costs" on the utilities books would be recoverable
from the ratepayers. According to the PUC, the costs would be subject to audit and review and thereafter a Commission
determination that specific amounts "were imprudently or unreasonably incurred." (562)
215
permitted the utility to recover through rate in-
creases certain energy costs that had been incurred
prior to TMI and had not previously been covered
through ratemaking. The amount the Board au-
thorized Jersey Central to recover was $17.9 mil-
lion in annual revenues, the same amount the
utility was having taken out of its rate base because
of the Board's decision on TMI-1 costs. (571) In
rendering these decisions, the New Jersey Board
said that it was aware of Jersey Central's serious
condition and would work toward its preservation
as an onsjoino; concern. (572)
Despite the favorable relief granted, New Jer-
sey's decision to remove Unit 1 costs from the rate
base led the banks to send GPU a letter, dated
April 9, 1980. In it, they said that "substantial
questions" remained as to the financial viability
of the utilities and that the $292 million credit
limit would not be raised until there was "greater
assurance" of viability, "including favorable regu-
latory action." (573)
On May 13, 1980, the New Jersey Utilities Board
granted Jersey Central additional, immediate rate
relief totaling $60 million. (574) Although this
relief was not directly attributable to TMI, it
nonetheless meant that Jersey Central would have
substantial additional revenues.
Two days later, on May 15, 1980, the banks sent
another letter to GPU. (575) This letter termed
the recent rate rulings as "significantly respon-
sive" " 5 but said "substantial questions" remained
as to the borrowers' ongoing financial viability.
The banks expressed particular concern over the
Pennsylvania PUC's May 9 decision to remove
Unit 1 costs from Met Ed's rate base.
Apart from its rate-making decisions, the New
Jersey Board of Public Utilities has been involved
in other issues of concern to Jersey Central. On
January 23, 1980, the Board decided to evaluate
several issues relating to Jersey Central's relative
fault for the accident, the regulatory consequences
of a finding of fault, the Board's legal authority
to impose those consequences and the implications
for the ratepayers and Jersey Central. (577) It
asked Jersey Central and other parties to set forth
their positions on these issues. (578)
The Board of Public Utilities also has been as-
sessing alternatives to Jersey Central's existing
operations. In early 1980 the Board retained
Arthur Young and Company to analyze a broad
range of options for Jersey Central, including the
transfer of part of Jersey Central's Service Terri-
tory or a State takeover of the utility. (579)
Regulatory proceedings also have been invoked
as part of the attempt by GPU and its subsidiaries
to modify the Pennsylvania-New Jersey-Maryland
(PJM) Interconnection Agreement. The GPU
companies have been purchasing replacement
power from the PJM pool on a split-savings ba-
sis each GPU utility had to pay a price halfway
between the cost to the selling utility and what it
would have cost the GPU utility to produce the
power through its own facilities. (580) In its
June 19, 1979 order authorizing rate adjustments
to reflect replacement power purchases, the Penn-
sylvania PUC directed Met Ed and PENELEC
to negotiate with other members of the PJM pool
for pricing not on a split-savings basis, but at
cost. (581)
Met Ed, PENELEC and Jersey Central were
unable to convince the other members of the pool
to agree to this. However, the members did pro-
pose to allow TMI-related purchases at cost plus
10 percent. (582) In October 1979, it was estimated
that if this change were put into effect for 1980,
the GPU utilities could save and the other mem-
bers of the PJM pool relinquish as much as $32
million. As part of this proposal, the pool-
member utilities were to petition their respective
state and city regulatory bodies for a finding that
the proposal was "in the public interest." (583)
Then the matter would be turned over to the Fed-
eral Energy Eegulatbry Commission, 146 which has
jurisdiction over the proposed PJM rate modifi-
cation. (That agency regulates all wholesale power
sales in the country.) (585)
As called for under the proposal, pool members
began taking steps to obtain approval of the pro-
posed PJM modification from utility regulators
in Virginia, Pennsylvania, Maryland, New Jersey
and the District of Columbia. Prompt approval
was not forthcoming. The D.C. Public Service
Commission gave indications that it might take
six months before reaching any decision on the
proposal, (586) which, if implemented, would have
resulted in increased costs to District customers.
(587)
Given the prospective delay, GPU decided in
late March to abandon efforts first to get state
regulatory approval of the cost plus 10 percent
arrangement negotiated by the other PJM pool
members. (588) GPU thus filed papers with the
Federal Energy Regulatory Commission stating
that it had "not been feasible to obtain an agree-
men